tempest-react-sdk 0.35.0 → 0.37.0

package/dist/vision/core/graph.cjs

@@ -0,0 +1,2 @@
+function e(e){return e===void 0?[]:e.map(e=>e.isTensor?e.shape.map(e=>typeof e==`number`&&Number.isInteger(e)&&e>0?e:null):[])}function t(e){if(e.length!==4)return null;let t=e[2],n=e[3];return t==null||n==null?null:[n,t]}function n(e){let n=e.graphShape===void 0?null:t(e.graphShape),r=e.requested;return n===null?r??e.fallback:(r!==void 0&&(r[0]!==n[0]||r[1]!==n[1])&&console.warn(`[ort-vision-sdk] The model declares a ${n[0]}x${n[1]} input; ignoring the requested ${r[0]}x${r[1]}, which ONNX Runtime would reject.`),n)}exports.declaredShapesFrom=e,exports.resolveInputSize=n,exports.spatialInputSize=t;
+//# sourceMappingURL=graph.cjs.map

package/dist/vision/core/graph.cjs.map

@@ -0,0 +1 @@
+{"version":3,"file":"graph.cjs","names":[],"sources":["../../../src/vision/core/graph.ts"],"sourcesContent":["/**\n * What the ONNX graph itself says about its inputs.\n *\n * The resolution a session must be fed at is a property of the exported model,\n * not of the configuration around it. Feeding a 640x640 tensor to a graph\n * exported at 224x224 makes ORT abort the run with\n * `Got invalid dimensions for input: images ... Got: 640 Expected: 224`, and the\n * caller has no way to see that coming from the outside — the number lives in\n * the file. So the SDK reads it from the graph and treats any configured size as\n * a fallback for when the graph leaves it open.\n */\n\nimport type * as ort from \"onnxruntime-web\";\n\n/**\n * One declared dimension: a number when the graph pins it, `null` when the\n * dimension is symbolic (dynamic).\n */\nexport type DeclaredDim = number | null;\n\n/** A declared input/output shape, dynamic axes appearing as `null`. */\nexport type DeclaredShape = readonly DeclaredDim[];\n\n/**\n * Convert ORT value metadata into declared shapes.\n *\n * @param metadata Metadata as reported by `InferenceSession.inputMetadata`, or\n *   `undefined` on ORT builds that predate it (added in onnxruntime 1.21).\n * @returns One shape per value, in declaration order. Non-tensor values and\n *   builds without metadata yield empty shapes, which read as \"nothing\n *   declared\" everywhere downstream.\n */\nexport function declaredShapesFrom(\n    metadata: readonly ort.InferenceSession.ValueMetadata[] | undefined,\n): readonly DeclaredShape[] {\n    if (metadata === undefined) return [];\n    return metadata.map((value) =>\n        value.isTensor\n            ? value.shape.map((dim) =>\n                  typeof dim === \"number\" && Number.isInteger(dim) && dim > 0 ? dim : null,\n              )\n            : [],\n    );\n}\n\n/**\n * Read the spatial input size out of a declared NCHW shape.\n *\n * @param shape The declared shape of the model's image input.\n * @returns `[width, height]` in pixels, or `null` when the shape is not 4D or\n *   leaves either spatial axis dynamic — in which case the model accepts more\n *   than one resolution and there is nothing to correct.\n */\nexport function spatialInputSize(shape: DeclaredShape): readonly [number, number] | null {\n    if (shape.length !== 4) return null;\n    const height = shape[2];\n    const width = shape[3];\n    if (height === null || height === undefined || width === null || width === undefined)\n        return null;\n    return [width, height];\n}\n\nexport interface ResolveInputSizeOptions {\n    /** Declared shape of the model's image input, from {@link declaredShapesFrom}. */\n    readonly graphShape?: DeclaredShape;\n    /** Size the caller asked for, if any. */\n    readonly requested?: readonly [number, number];\n    /** Size to use when neither the graph nor the caller pins one. */\n    readonly fallback: readonly [number, number];\n}\n\n/**\n * Decide the input size a task will preprocess to.\n *\n * Precedence is graph → caller → fallback. The graph wins over an explicit\n * `inputSize` because a static shape is not a preference, it is what ORT will\n * accept: honoring the caller there would only turn a fixable mismatch into a\n * failed run. A disagreement is a configuration bug in the caller, so it is\n * reported through `console.warn` instead of being swallowed.\n *\n * @param options Graph shape, requested size and per-task fallback.\n * @returns The `[width, height]` to preprocess to.\n */\nexport function resolveInputSize(options: ResolveInputSizeOptions): readonly [number, number] {\n    const graph = options.graphShape === undefined ? null : spatialInputSize(options.graphShape);\n    const requested = options.requested;\n    if (graph === null) return requested ?? options.fallback;\n    if (requested !== undefined && (requested[0] !== graph[0] || requested[1] !== graph[1])) {\n        console.warn(\n            `[ort-vision-sdk] The model declares a ${graph[0]}x${graph[1]} input; ` +\n                `ignoring the requested ${requested[0]}x${requested[1]}, which ONNX Runtime would reject.`,\n        );\n    }\n    return graph;\n}\n"],"mappings":"AAgCA,SAAgB,EACZ,EACwB,CAExB,OADI,IAAa,IAAA,GAAkB,CAAC,EAC7B,EAAS,IAAK,GACjB,EAAM,SACA,EAAM,MAAM,IAAK,GACb,OAAO,GAAQ,UAAY,OAAO,UAAU,CAAG,GAAK,EAAM,EAAI,EAAM,IACxE,EACA,CAAC,CACX,CACJ,CAUA,SAAgB,EAAiB,EAAwD,CACrF,GAAI,EAAM,SAAW,EAAG,OAAO,KAC/B,IAAM,EAAS,EAAM,GACf,EAAQ,EAAM,GAGpB,OAFI,GAAW,MAAgC,GAAU,KAC9C,KACJ,CAAC,EAAO,CAAM,CACzB,CAuBA,SAAgB,EAAiB,EAA6D,CAC1F,IAAM,EAAQ,EAAQ,aAAe,IAAA,GAAY,KAAO,EAAiB,EAAQ,UAAU,EACrF,EAAY,EAAQ,UAQ1B,OAPI,IAAU,KAAa,GAAa,EAAQ,UAC5C,IAAc,IAAA,KAAc,EAAU,KAAO,EAAM,IAAM,EAAU,KAAO,EAAM,KAChF,QAAQ,KACJ,yCAAyC,EAAM,GAAG,GAAG,EAAM,GAAG,iCAChC,EAAU,GAAG,GAAG,EAAU,GAAG,mCAC/D,EAEG,EACX"}

package/dist/vision/core/graph.js

@@ -0,0 +1,17 @@
+//#region src/vision/core/graph.ts
+function e(e) {
+	return e === void 0 ? [] : e.map((e) => e.isTensor ? e.shape.map((e) => typeof e == "number" && Number.isInteger(e) && e > 0 ? e : null) : []);
+}
+function t(e) {
+	if (e.length !== 4) return null;
+	let t = e[2], n = e[3];
+	return t == null || n == null ? null : [n, t];
+}
+function n(e) {
+	let n = e.graphShape === void 0 ? null : t(e.graphShape), r = e.requested;
+	return n === null ? r ?? e.fallback : (r !== void 0 && (r[0] !== n[0] || r[1] !== n[1]) && console.warn(`[ort-vision-sdk] The model declares a ${n[0]}x${n[1]} input; ignoring the requested ${r[0]}x${r[1]}, which ONNX Runtime would reject.`), n);
+}
+//#endregion
+export { e as declaredShapesFrom, n as resolveInputSize, t as spatialInputSize };
+
+//# sourceMappingURL=graph.js.map

package/dist/vision/core/graph.js.map

@@ -0,0 +1 @@
+{"version":3,"file":"graph.js","names":[],"sources":["../../../src/vision/core/graph.ts"],"sourcesContent":["/**\n * What the ONNX graph itself says about its inputs.\n *\n * The resolution a session must be fed at is a property of the exported model,\n * not of the configuration around it. Feeding a 640x640 tensor to a graph\n * exported at 224x224 makes ORT abort the run with\n * `Got invalid dimensions for input: images ... Got: 640 Expected: 224`, and the\n * caller has no way to see that coming from the outside — the number lives in\n * the file. So the SDK reads it from the graph and treats any configured size as\n * a fallback for when the graph leaves it open.\n */\n\nimport type * as ort from \"onnxruntime-web\";\n\n/**\n * One declared dimension: a number when the graph pins it, `null` when the\n * dimension is symbolic (dynamic).\n */\nexport type DeclaredDim = number | null;\n\n/** A declared input/output shape, dynamic axes appearing as `null`. */\nexport type DeclaredShape = readonly DeclaredDim[];\n\n/**\n * Convert ORT value metadata into declared shapes.\n *\n * @param metadata Metadata as reported by `InferenceSession.inputMetadata`, or\n *   `undefined` on ORT builds that predate it (added in onnxruntime 1.21).\n * @returns One shape per value, in declaration order. Non-tensor values and\n *   builds without metadata yield empty shapes, which read as \"nothing\n *   declared\" everywhere downstream.\n */\nexport function declaredShapesFrom(\n    metadata: readonly ort.InferenceSession.ValueMetadata[] | undefined,\n): readonly DeclaredShape[] {\n    if (metadata === undefined) return [];\n    return metadata.map((value) =>\n        value.isTensor\n            ? value.shape.map((dim) =>\n                  typeof dim === \"number\" && Number.isInteger(dim) && dim > 0 ? dim : null,\n              )\n            : [],\n    );\n}\n\n/**\n * Read the spatial input size out of a declared NCHW shape.\n *\n * @param shape The declared shape of the model's image input.\n * @returns `[width, height]` in pixels, or `null` when the shape is not 4D or\n *   leaves either spatial axis dynamic — in which case the model accepts more\n *   than one resolution and there is nothing to correct.\n */\nexport function spatialInputSize(shape: DeclaredShape): readonly [number, number] | null {\n    if (shape.length !== 4) return null;\n    const height = shape[2];\n    const width = shape[3];\n    if (height === null || height === undefined || width === null || width === undefined)\n        return null;\n    return [width, height];\n}\n\nexport interface ResolveInputSizeOptions {\n    /** Declared shape of the model's image input, from {@link declaredShapesFrom}. */\n    readonly graphShape?: DeclaredShape;\n    /** Size the caller asked for, if any. */\n    readonly requested?: readonly [number, number];\n    /** Size to use when neither the graph nor the caller pins one. */\n    readonly fallback: readonly [number, number];\n}\n\n/**\n * Decide the input size a task will preprocess to.\n *\n * Precedence is graph → caller → fallback. The graph wins over an explicit\n * `inputSize` because a static shape is not a preference, it is what ORT will\n * accept: honoring the caller there would only turn a fixable mismatch into a\n * failed run. A disagreement is a configuration bug in the caller, so it is\n * reported through `console.warn` instead of being swallowed.\n *\n * @param options Graph shape, requested size and per-task fallback.\n * @returns The `[width, height]` to preprocess to.\n */\nexport function resolveInputSize(options: ResolveInputSizeOptions): readonly [number, number] {\n    const graph = options.graphShape === undefined ? null : spatialInputSize(options.graphShape);\n    const requested = options.requested;\n    if (graph === null) return requested ?? options.fallback;\n    if (requested !== undefined && (requested[0] !== graph[0] || requested[1] !== graph[1])) {\n        console.warn(\n            `[ort-vision-sdk] The model declares a ${graph[0]}x${graph[1]} input; ` +\n                `ignoring the requested ${requested[0]}x${requested[1]}, which ONNX Runtime would reject.`,\n        );\n    }\n    return graph;\n}\n"],"mappings":";AAgCA,SAAgB,EACZ,GACwB;CAExB,OADI,MAAa,KAAA,IAAkB,CAAC,IAC7B,EAAS,KAAK,MACjB,EAAM,WACA,EAAM,MAAM,KAAK,MACb,OAAO,KAAQ,YAAY,OAAO,UAAU,CAAG,KAAK,IAAM,IAAI,IAAM,IACxE,IACA,CAAC,CACX;AACJ;AAUA,SAAgB,EAAiB,GAAwD;CACrF,IAAI,EAAM,WAAW,GAAG,OAAO;CAC/B,IAAM,IAAS,EAAM,IACf,IAAQ,EAAM;CAGpB,OAFI,KAAW,QAAgC,KAAU,OAC9C,OACJ,CAAC,GAAO,CAAM;AACzB;AAuBA,SAAgB,EAAiB,GAA6D;CAC1F,IAAM,IAAQ,EAAQ,eAAe,KAAA,IAAY,OAAO,EAAiB,EAAQ,UAAU,GACrF,IAAY,EAAQ;CAQ1B,OAPI,MAAU,OAAa,KAAa,EAAQ,YAC5C,MAAc,KAAA,MAAc,EAAU,OAAO,EAAM,MAAM,EAAU,OAAO,EAAM,OAChF,QAAQ,KACJ,yCAAyC,EAAM,GAAG,GAAG,EAAM,GAAG,iCAChC,EAAU,GAAG,GAAG,EAAU,GAAG,mCAC/D,GAEG;AACX"}

package/dist/vision/core/session.cjs

@@ -1,2 +1,2 @@
-const e=require("../../_virtual/_rolldown/runtime.cjs"),t=require("./exceptions.cjs"),n=require("./providers.cjs");let r=require("onnxruntime-web");r=e.__toESM(r,1);var i=class e{_session;providers;constructor(e,t){this._session=e,this.providers=t}static async create(i,a={}){let o=n.resolveProviders(a.providers),s={...a.sessionOptions??{},executionProviders:o},c;try{c=(typeof i==`string`||i instanceof Uint8Array,await r.InferenceSession.create(i,s))}catch(e){throw new t.ModelLoadError(`Failed to load ONNX model: ${e.message}`,{cause:e})}return new e(c,o)}get inputNames(){return this._session.inputNames}get inputName(){let e=this._session.inputNames[0];if(e===void 0)throw new t.InferenceError(`Model has no inputs.`);return e}get outputNames(){return this._session.outputNames}get raw(){return this._session}async run(e){try{return await this._session.run(e)}catch(e){throw new t.InferenceError(`Inference failed: ${e.message}`,{cause:e})}}};exports.OrtSession=i;
+const e=require("../../_virtual/_rolldown/runtime.cjs"),t=require("./exceptions.cjs"),n=require("./graph.cjs"),r=require("./providers.cjs");let i=require("onnxruntime-web");i=e.__toESM(i,1);var a=class e{_session;providers;constructor(e,t){this._session=e,this.providers=t}static async create(n,a={}){let o=r.resolveProviders(a.providers),s={...a.sessionOptions??{},executionProviders:o},c;try{c=(typeof n==`string`||n instanceof Uint8Array,await i.InferenceSession.create(n,s))}catch(e){throw new t.ModelLoadError(`Failed to load ONNX model: ${e.message}`,{cause:e})}return new e(c,o)}get inputNames(){return this._session.inputNames}get inputName(){let e=this._session.inputNames[0];if(e===void 0)throw new t.InferenceError(`Model has no inputs.`);return e}get outputNames(){return this._session.outputNames}get inputShapes(){return n.declaredShapesFrom(this._session.inputMetadata)}get inputShape(){return this.inputShapes[0]??[]}async release(){await this._session.release().catch(()=>void 0)}get raw(){return this._session}async run(e){try{return await this._session.run(e)}catch(e){throw new t.InferenceError(`Inference failed: ${e.message}`,{cause:e})}}};exports.OrtSession=a;
 //# sourceMappingURL=session.cjs.map

package/dist/vision/core/session.cjs.map

@@ -1 +1 @@
-{"version":3,"file":"session.cjs","names":[],"sources":["../../../src/vision/core/session.ts"],"sourcesContent":["/**\n * Thin wrapper around `onnxruntime-web` `InferenceSession` with typed metadata.\n */\n\nimport type * as ort from \"onnxruntime-web\";\nimport * as ortRuntime from \"onnxruntime-web\";\n\nimport { InferenceError, ModelLoadError } from \"./exceptions\";\nimport { resolveProviders } from \"./providers\";\n\n/** Anything `InferenceSession.create` accepts. */\nexport type ModelSource = string | ArrayBufferLike | Uint8Array;\n\nexport interface OrtSessionOptions {\n    /** Execution providers in preference order. `undefined` uses {@link DEFAULT_PROVIDERS}. */\n    readonly providers?: readonly string[];\n    /** Optional ORT session options forwarded to `InferenceSession.create`. */\n    readonly sessionOptions?: ort.InferenceSession.SessionOptions;\n}\n\n/**\n * Wrap an ONNX Runtime Web `InferenceSession` with convenient metadata access.\n *\n * The wrapper exposes input/output names, manages execution-provider\n * selection, and provides a typed {@link OrtSession.run} method.\n */\nexport class OrtSession {\n    private constructor(\n        private readonly _session: ort.InferenceSession,\n        public readonly providers: readonly string[],\n    ) {}\n\n    /**\n     * Load an ONNX model into an ORT inference session.\n     *\n     * @param model Either a URL string fetched by ORT, or a `Uint8Array`/`ArrayBuffer` containing the model bytes.\n     * @param options Provider list and pass-through `SessionOptions`.\n     * @throws {@link ModelLoadError} if the model cannot be loaded.\n     */\n    static async create(model: ModelSource, options: OrtSessionOptions = {}): Promise<OrtSession> {\n        const providers = resolveProviders(options.providers);\n        const sessionOptions: ort.InferenceSession.SessionOptions = {\n            ...(options.sessionOptions ?? {}),\n            executionProviders:\n                providers as ort.InferenceSession.SessionOptions[\"executionProviders\"],\n        };\n\n        let session: ort.InferenceSession;\n        try {\n            if (typeof model === \"string\") {\n                session = await ortRuntime.InferenceSession.create(model, sessionOptions);\n            } else if (model instanceof Uint8Array) {\n                session = await ortRuntime.InferenceSession.create(model, sessionOptions);\n            } else {\n                session = await ortRuntime.InferenceSession.create(\n                    model as ArrayBuffer,\n                    sessionOptions,\n                );\n            }\n        } catch (err) {\n            throw new ModelLoadError(`Failed to load ONNX model: ${(err as Error).message}`, {\n                cause: err,\n            });\n        }\n\n        return new OrtSession(session, providers);\n    }\n\n    /** Names of the model's inputs, in declaration order. */\n    get inputNames(): readonly string[] {\n        return this._session.inputNames;\n    }\n\n    /** Name of the first (and usually only) input. */\n    get inputName(): string {\n        const name = this._session.inputNames[0];\n        if (name === undefined) {\n            throw new InferenceError(\"Model has no inputs.\");\n        }\n        return name;\n    }\n\n    /** Names of the model's outputs, in declaration order. */\n    get outputNames(): readonly string[] {\n        return this._session.outputNames;\n    }\n\n    /** The underlying `onnxruntime-web` session, for advanced use cases. */\n    get raw(): ort.InferenceSession {\n        return this._session;\n    }\n\n    /**\n     * Run inference and return all outputs.\n     *\n     * @param feeds Map of input name to `ort.Tensor`. Keys must match {@link inputNames}.\n     * @throws {@link InferenceError} if ORT raises any error during execution.\n     */\n    async run(feeds: Record<string, ort.Tensor>): Promise<Record<string, ort.Tensor>> {\n        try {\n            const result = await this._session.run(feeds);\n            return result as Record<string, ort.Tensor>;\n        } catch (err) {\n            throw new InferenceError(`Inference failed: ${(err as Error).message}`, { cause: err });\n        }\n    }\n}\n"],"mappings":"qKA0BA,IAAa,EAAb,MAAa,CAAW,CAEC,SACD,UAFpB,YACI,EACA,EACF,CAFmB,KAAA,SAAA,EACD,KAAA,UAAA,CACjB,CASH,aAAa,OAAO,EAAoB,EAA6B,CAAC,EAAwB,CAC1F,IAAM,EAAY,EAAA,iBAAiB,EAAQ,SAAS,EAC9C,EAAsD,CACxD,GAAI,EAAQ,gBAAkB,CAAC,EAC/B,mBACI,CACR,EAEI,EACJ,GAAI,CACA,AAKI,GALA,OAAO,GAAU,UAEV,aAAiB,WADd,MAAM,EAAW,iBAAiB,OAAO,EAAO,CAAc,EAShF,OAAS,EAAK,CACV,MAAM,IAAI,EAAA,eAAe,8BAA+B,EAAc,UAAW,CAC7E,MAAO,CACX,CAAC,CACL,CAEA,OAAO,IAAI,EAAW,EAAS,CAAS,CAC5C,CAGA,IAAI,YAAgC,CAChC,OAAO,KAAK,SAAS,UACzB,CAGA,IAAI,WAAoB,CACpB,IAAM,EAAO,KAAK,SAAS,WAAW,GACtC,GAAI,IAAS,IAAA,GACT,MAAM,IAAI,EAAA,eAAe,sBAAsB,EAEnD,OAAO,CACX,CAGA,IAAI,aAAiC,CACjC,OAAO,KAAK,SAAS,WACzB,CAGA,IAAI,KAA4B,CAC5B,OAAO,KAAK,QAChB,CAQA,MAAM,IAAI,EAAwE,CAC9E,GAAI,CAEA,OAAO,MADc,KAAK,SAAS,IAAI,CAAK,CAEhD,OAAS,EAAK,CACV,MAAM,IAAI,EAAA,eAAe,qBAAsB,EAAc,UAAW,CAAE,MAAO,CAAI,CAAC,CAC1F,CACJ,CACJ"}
+{"version":3,"file":"session.cjs","names":[],"sources":["../../../src/vision/core/session.ts"],"sourcesContent":["/**\n * Thin wrapper around `onnxruntime-web` `InferenceSession` with typed metadata.\n */\n\nimport type * as ort from \"onnxruntime-web\";\nimport * as ortRuntime from \"onnxruntime-web\";\n\nimport { InferenceError, ModelLoadError } from \"./exceptions\";\nimport { type DeclaredShape, declaredShapesFrom } from \"./graph\";\nimport { resolveProviders } from \"./providers\";\n\n/** Anything `InferenceSession.create` accepts. */\nexport type ModelSource = string | ArrayBufferLike | Uint8Array;\n\nexport interface OrtSessionOptions {\n    /** Execution providers in preference order. `undefined` uses {@link DEFAULT_PROVIDERS}. */\n    readonly providers?: readonly string[];\n    /** Optional ORT session options forwarded to `InferenceSession.create`. */\n    readonly sessionOptions?: ort.InferenceSession.SessionOptions;\n}\n\n/**\n * Wrap an ONNX Runtime Web `InferenceSession` with convenient metadata access.\n *\n * The wrapper exposes input/output names and the shapes the graph declares,\n * manages execution-provider selection, provides a typed {@link OrtSession.run}\n * method, and releases the native session through {@link OrtSession.release}.\n */\nexport class OrtSession {\n    private constructor(\n        private readonly _session: ort.InferenceSession,\n        public readonly providers: readonly string[],\n    ) {}\n\n    /**\n     * Load an ONNX model into an ORT inference session.\n     *\n     * @param model Either a URL string fetched by ORT, or a `Uint8Array`/`ArrayBuffer` containing the model bytes.\n     * @param options Provider list and pass-through `SessionOptions`.\n     * @throws {@link ModelLoadError} if the model cannot be loaded.\n     */\n    static async create(model: ModelSource, options: OrtSessionOptions = {}): Promise<OrtSession> {\n        const providers = resolveProviders(options.providers);\n        const sessionOptions: ort.InferenceSession.SessionOptions = {\n            ...(options.sessionOptions ?? {}),\n            executionProviders:\n                providers as ort.InferenceSession.SessionOptions[\"executionProviders\"],\n        };\n\n        let session: ort.InferenceSession;\n        try {\n            if (typeof model === \"string\") {\n                session = await ortRuntime.InferenceSession.create(model, sessionOptions);\n            } else if (model instanceof Uint8Array) {\n                session = await ortRuntime.InferenceSession.create(model, sessionOptions);\n            } else {\n                session = await ortRuntime.InferenceSession.create(\n                    model as ArrayBuffer,\n                    sessionOptions,\n                );\n            }\n        } catch (err) {\n            throw new ModelLoadError(`Failed to load ONNX model: ${(err as Error).message}`, {\n                cause: err,\n            });\n        }\n\n        return new OrtSession(session, providers);\n    }\n\n    /** Names of the model's inputs, in declaration order. */\n    get inputNames(): readonly string[] {\n        return this._session.inputNames;\n    }\n\n    /** Name of the first (and usually only) input. */\n    get inputName(): string {\n        const name = this._session.inputNames[0];\n        if (name === undefined) {\n            throw new InferenceError(\"Model has no inputs.\");\n        }\n        return name;\n    }\n\n    /** Names of the model's outputs, in declaration order. */\n    get outputNames(): readonly string[] {\n        return this._session.outputNames;\n    }\n\n    /**\n     * Shapes the graph declares for its inputs, in declaration order.\n     *\n     * Dynamic (symbolic) axes appear as `null`. Empty shapes mean the runtime\n     * reported no metadata — either a non-tensor input, or an `onnxruntime-web`\n     * older than 1.21, which predates input metadata.\n     */\n    get inputShapes(): readonly DeclaredShape[] {\n        return declaredShapesFrom(\n            this._session.inputMetadata as\n                readonly ort.InferenceSession.ValueMetadata[] | undefined,\n        );\n    }\n\n    /**\n     * Shape the graph declares for its first input, dynamic axes as `null`.\n     *\n     * Empty when the runtime reports no metadata for it.\n     */\n    get inputShape(): DeclaredShape {\n        return this.inputShapes[0] ?? [];\n    }\n\n    /**\n     * Release the native session and free its memory.\n     *\n     * Call it when a session is discarded while the page lives on — rebuilding a\n     * task at a different input size, swapping in a newer model. A failure from\n     * the runtime is ignored: a session being torn down has nothing left to fail\n     * at, and the caller is already moving on.\n     */\n    async release(): Promise<void> {\n        await this._session.release().catch(() => undefined);\n    }\n\n    /** The underlying `onnxruntime-web` session, for advanced use cases. */\n    get raw(): ort.InferenceSession {\n        return this._session;\n    }\n\n    /**\n     * Run inference and return all outputs.\n     *\n     * @param feeds Map of input name to `ort.Tensor`. Keys must match {@link inputNames}.\n     * @throws {@link InferenceError} if ORT raises any error during execution.\n     */\n    async run(feeds: Record<string, ort.Tensor>): Promise<Record<string, ort.Tensor>> {\n        try {\n            const result = await this._session.run(feeds);\n            return result as Record<string, ort.Tensor>;\n        } catch (err) {\n            throw new InferenceError(`Inference failed: ${(err as Error).message}`, { cause: err });\n        }\n    }\n}\n"],"mappings":"8LA4BA,IAAa,EAAb,MAAa,CAAW,CAEC,SACD,UAFpB,YACI,EACA,EACF,CAFmB,KAAA,SAAA,EACD,KAAA,UAAA,CACjB,CASH,aAAa,OAAO,EAAoB,EAA6B,CAAC,EAAwB,CAC1F,IAAM,EAAY,EAAA,iBAAiB,EAAQ,SAAS,EAC9C,EAAsD,CACxD,GAAI,EAAQ,gBAAkB,CAAC,EAC/B,mBACI,CACR,EAEI,EACJ,GAAI,CACA,AAKI,GALA,OAAO,GAAU,UAEV,aAAiB,WADd,MAAM,EAAW,iBAAiB,OAAO,EAAO,CAAc,EAShF,OAAS,EAAK,CACV,MAAM,IAAI,EAAA,eAAe,8BAA+B,EAAc,UAAW,CAC7E,MAAO,CACX,CAAC,CACL,CAEA,OAAO,IAAI,EAAW,EAAS,CAAS,CAC5C,CAGA,IAAI,YAAgC,CAChC,OAAO,KAAK,SAAS,UACzB,CAGA,IAAI,WAAoB,CACpB,IAAM,EAAO,KAAK,SAAS,WAAW,GACtC,GAAI,IAAS,IAAA,GACT,MAAM,IAAI,EAAA,eAAe,sBAAsB,EAEnD,OAAO,CACX,CAGA,IAAI,aAAiC,CACjC,OAAO,KAAK,SAAS,WACzB,CASA,IAAI,aAAwC,CACxC,OAAO,EAAA,mBACH,KAAK,SAAS,aAElB,CACJ,CAOA,IAAI,YAA4B,CAC5B,OAAO,KAAK,YAAY,IAAM,CAAC,CACnC,CAUA,MAAM,SAAyB,CAC3B,MAAM,KAAK,SAAS,QAAQ,CAAC,CAAC,UAAY,IAAA,EAAS,CACvD,CAGA,IAAI,KAA4B,CAC5B,OAAO,KAAK,QAChB,CAQA,MAAM,IAAI,EAAwE,CAC9E,GAAI,CAEA,OAAO,MADc,KAAK,SAAS,IAAI,CAAK,CAEhD,OAAS,EAAK,CACV,MAAM,IAAI,EAAA,eAAe,qBAAsB,EAAc,UAAW,CAAE,MAAO,CAAI,CAAC,CAC1F,CACJ,CACJ"}

package/dist/vision/core/session.js

@@ -1,24 +1,25 @@
 import { InferenceError as e, ModelLoadError as t } from "./exceptions.js";
-import { resolveProviders as n } from "./providers.js";
-import * as r from "onnxruntime-web";
+import { declaredShapesFrom as n } from "./graph.js";
+import { resolveProviders as r } from "./providers.js";
+import * as i from "onnxruntime-web";
 //#region src/vision/core/session.ts
-var i = class i {
+var a = class a {
 	_session;
 	providers;
 	constructor(e, t) {
 		this._session = e, this.providers = t;
 	}
-	static async create(e, a = {}) {
-		let o = n(a.providers), s = {
-			...a.sessionOptions ?? {},
+	static async create(e, n = {}) {
+		let o = r(n.providers), s = {
+			...n.sessionOptions ?? {},
 			executionProviders: o
 		}, c;
 		try {
-			c = (typeof e == "string" || e instanceof Uint8Array, await r.InferenceSession.create(e, s));
+			c = (typeof e == "string" || e instanceof Uint8Array, await i.InferenceSession.create(e, s));
 		} catch (e) {
 			throw new t(`Failed to load ONNX model: ${e.message}`, { cause: e });
 		}
-		return new i(c, o);
+		return new a(c, o);
 	}
 	get inputNames() {
 		return this._session.inputNames;
@@ -31,6 +32,15 @@ var i = class i {
 	get outputNames() {
 		return this._session.outputNames;
 	}
+	get inputShapes() {
+		return n(this._session.inputMetadata);
+	}
+	get inputShape() {
+		return this.inputShapes[0] ?? [];
+	}
+	async release() {
+		await this._session.release().catch(() => void 0);
+	}
 	get raw() {
 		return this._session;
 	}
@@ -43,6 +53,6 @@ var i = class i {
 	}
 };
 //#endregion
-export { i as OrtSession };
+export { a as OrtSession };
 
 //# sourceMappingURL=session.js.map

package/dist/vision/core/session.js.map

@@ -1 +1 @@
-{"version":3,"file":"session.js","names":[],"sources":["../../../src/vision/core/session.ts"],"sourcesContent":["/**\n * Thin wrapper around `onnxruntime-web` `InferenceSession` with typed metadata.\n */\n\nimport type * as ort from \"onnxruntime-web\";\nimport * as ortRuntime from \"onnxruntime-web\";\n\nimport { InferenceError, ModelLoadError } from \"./exceptions\";\nimport { resolveProviders } from \"./providers\";\n\n/** Anything `InferenceSession.create` accepts. */\nexport type ModelSource = string | ArrayBufferLike | Uint8Array;\n\nexport interface OrtSessionOptions {\n    /** Execution providers in preference order. `undefined` uses {@link DEFAULT_PROVIDERS}. */\n    readonly providers?: readonly string[];\n    /** Optional ORT session options forwarded to `InferenceSession.create`. */\n    readonly sessionOptions?: ort.InferenceSession.SessionOptions;\n}\n\n/**\n * Wrap an ONNX Runtime Web `InferenceSession` with convenient metadata access.\n *\n * The wrapper exposes input/output names, manages execution-provider\n * selection, and provides a typed {@link OrtSession.run} method.\n */\nexport class OrtSession {\n    private constructor(\n        private readonly _session: ort.InferenceSession,\n        public readonly providers: readonly string[],\n    ) {}\n\n    /**\n     * Load an ONNX model into an ORT inference session.\n     *\n     * @param model Either a URL string fetched by ORT, or a `Uint8Array`/`ArrayBuffer` containing the model bytes.\n     * @param options Provider list and pass-through `SessionOptions`.\n     * @throws {@link ModelLoadError} if the model cannot be loaded.\n     */\n    static async create(model: ModelSource, options: OrtSessionOptions = {}): Promise<OrtSession> {\n        const providers = resolveProviders(options.providers);\n        const sessionOptions: ort.InferenceSession.SessionOptions = {\n            ...(options.sessionOptions ?? {}),\n            executionProviders:\n                providers as ort.InferenceSession.SessionOptions[\"executionProviders\"],\n        };\n\n        let session: ort.InferenceSession;\n        try {\n            if (typeof model === \"string\") {\n                session = await ortRuntime.InferenceSession.create(model, sessionOptions);\n            } else if (model instanceof Uint8Array) {\n                session = await ortRuntime.InferenceSession.create(model, sessionOptions);\n            } else {\n                session = await ortRuntime.InferenceSession.create(\n                    model as ArrayBuffer,\n                    sessionOptions,\n                );\n            }\n        } catch (err) {\n            throw new ModelLoadError(`Failed to load ONNX model: ${(err as Error).message}`, {\n                cause: err,\n            });\n        }\n\n        return new OrtSession(session, providers);\n    }\n\n    /** Names of the model's inputs, in declaration order. */\n    get inputNames(): readonly string[] {\n        return this._session.inputNames;\n    }\n\n    /** Name of the first (and usually only) input. */\n    get inputName(): string {\n        const name = this._session.inputNames[0];\n        if (name === undefined) {\n            throw new InferenceError(\"Model has no inputs.\");\n        }\n        return name;\n    }\n\n    /** Names of the model's outputs, in declaration order. */\n    get outputNames(): readonly string[] {\n        return this._session.outputNames;\n    }\n\n    /** The underlying `onnxruntime-web` session, for advanced use cases. */\n    get raw(): ort.InferenceSession {\n        return this._session;\n    }\n\n    /**\n     * Run inference and return all outputs.\n     *\n     * @param feeds Map of input name to `ort.Tensor`. Keys must match {@link inputNames}.\n     * @throws {@link InferenceError} if ORT raises any error during execution.\n     */\n    async run(feeds: Record<string, ort.Tensor>): Promise<Record<string, ort.Tensor>> {\n        try {\n            const result = await this._session.run(feeds);\n            return result as Record<string, ort.Tensor>;\n        } catch (err) {\n            throw new InferenceError(`Inference failed: ${(err as Error).message}`, { cause: err });\n        }\n    }\n}\n"],"mappings":";;;;AA0BA,IAAa,IAAb,MAAa,EAAW;CAEC;CACD;CAFpB,YACI,GACA,GACF;EADkB,AADC,KAAA,WAAA,GACD,KAAA,YAAA;CACjB;CASH,aAAa,OAAO,GAAoB,IAA6B,CAAC,GAAwB;EAC1F,IAAM,IAAY,EAAiB,EAAQ,SAAS,GAC9C,IAAsD;GACxD,GAAI,EAAQ,kBAAkB,CAAC;GAC/B,oBACI;EACR,GAEI;EACJ,IAAI;GACA,AAKI,KALA,OAAO,KAAU,YAEV,aAAiB,YADd,MAAM,EAAW,iBAAiB,OAAO,GAAO,CAAc;EAShF,SAAS,GAAK;GACV,MAAM,IAAI,EAAe,8BAA+B,EAAc,WAAW,EAC7E,OAAO,EACX,CAAC;EACL;EAEA,OAAO,IAAI,EAAW,GAAS,CAAS;CAC5C;CAGA,IAAI,aAAgC;EAChC,OAAO,KAAK,SAAS;CACzB;CAGA,IAAI,YAAoB;EACpB,IAAM,IAAO,KAAK,SAAS,WAAW;EACtC,IAAI,MAAS,KAAA,GACT,MAAM,IAAI,EAAe,sBAAsB;EAEnD,OAAO;CACX;CAGA,IAAI,cAAiC;EACjC,OAAO,KAAK,SAAS;CACzB;CAGA,IAAI,MAA4B;EAC5B,OAAO,KAAK;CAChB;CAQA,MAAM,IAAI,GAAwE;EAC9E,IAAI;GAEA,OAAO,MADc,KAAK,SAAS,IAAI,CAAK;EAEhD,SAAS,GAAK;GACV,MAAM,IAAI,EAAe,qBAAsB,EAAc,WAAW,EAAE,OAAO,EAAI,CAAC;EAC1F;CACJ;AACJ"}
+{"version":3,"file":"session.js","names":[],"sources":["../../../src/vision/core/session.ts"],"sourcesContent":["/**\n * Thin wrapper around `onnxruntime-web` `InferenceSession` with typed metadata.\n */\n\nimport type * as ort from \"onnxruntime-web\";\nimport * as ortRuntime from \"onnxruntime-web\";\n\nimport { InferenceError, ModelLoadError } from \"./exceptions\";\nimport { type DeclaredShape, declaredShapesFrom } from \"./graph\";\nimport { resolveProviders } from \"./providers\";\n\n/** Anything `InferenceSession.create` accepts. */\nexport type ModelSource = string | ArrayBufferLike | Uint8Array;\n\nexport interface OrtSessionOptions {\n    /** Execution providers in preference order. `undefined` uses {@link DEFAULT_PROVIDERS}. */\n    readonly providers?: readonly string[];\n    /** Optional ORT session options forwarded to `InferenceSession.create`. */\n    readonly sessionOptions?: ort.InferenceSession.SessionOptions;\n}\n\n/**\n * Wrap an ONNX Runtime Web `InferenceSession` with convenient metadata access.\n *\n * The wrapper exposes input/output names and the shapes the graph declares,\n * manages execution-provider selection, provides a typed {@link OrtSession.run}\n * method, and releases the native session through {@link OrtSession.release}.\n */\nexport class OrtSession {\n    private constructor(\n        private readonly _session: ort.InferenceSession,\n        public readonly providers: readonly string[],\n    ) {}\n\n    /**\n     * Load an ONNX model into an ORT inference session.\n     *\n     * @param model Either a URL string fetched by ORT, or a `Uint8Array`/`ArrayBuffer` containing the model bytes.\n     * @param options Provider list and pass-through `SessionOptions`.\n     * @throws {@link ModelLoadError} if the model cannot be loaded.\n     */\n    static async create(model: ModelSource, options: OrtSessionOptions = {}): Promise<OrtSession> {\n        const providers = resolveProviders(options.providers);\n        const sessionOptions: ort.InferenceSession.SessionOptions = {\n            ...(options.sessionOptions ?? {}),\n            executionProviders:\n                providers as ort.InferenceSession.SessionOptions[\"executionProviders\"],\n        };\n\n        let session: ort.InferenceSession;\n        try {\n            if (typeof model === \"string\") {\n                session = await ortRuntime.InferenceSession.create(model, sessionOptions);\n            } else if (model instanceof Uint8Array) {\n                session = await ortRuntime.InferenceSession.create(model, sessionOptions);\n            } else {\n                session = await ortRuntime.InferenceSession.create(\n                    model as ArrayBuffer,\n                    sessionOptions,\n                );\n            }\n        } catch (err) {\n            throw new ModelLoadError(`Failed to load ONNX model: ${(err as Error).message}`, {\n                cause: err,\n            });\n        }\n\n        return new OrtSession(session, providers);\n    }\n\n    /** Names of the model's inputs, in declaration order. */\n    get inputNames(): readonly string[] {\n        return this._session.inputNames;\n    }\n\n    /** Name of the first (and usually only) input. */\n    get inputName(): string {\n        const name = this._session.inputNames[0];\n        if (name === undefined) {\n            throw new InferenceError(\"Model has no inputs.\");\n        }\n        return name;\n    }\n\n    /** Names of the model's outputs, in declaration order. */\n    get outputNames(): readonly string[] {\n        return this._session.outputNames;\n    }\n\n    /**\n     * Shapes the graph declares for its inputs, in declaration order.\n     *\n     * Dynamic (symbolic) axes appear as `null`. Empty shapes mean the runtime\n     * reported no metadata — either a non-tensor input, or an `onnxruntime-web`\n     * older than 1.21, which predates input metadata.\n     */\n    get inputShapes(): readonly DeclaredShape[] {\n        return declaredShapesFrom(\n            this._session.inputMetadata as\n                readonly ort.InferenceSession.ValueMetadata[] | undefined,\n        );\n    }\n\n    /**\n     * Shape the graph declares for its first input, dynamic axes as `null`.\n     *\n     * Empty when the runtime reports no metadata for it.\n     */\n    get inputShape(): DeclaredShape {\n        return this.inputShapes[0] ?? [];\n    }\n\n    /**\n     * Release the native session and free its memory.\n     *\n     * Call it when a session is discarded while the page lives on — rebuilding a\n     * task at a different input size, swapping in a newer model. A failure from\n     * the runtime is ignored: a session being torn down has nothing left to fail\n     * at, and the caller is already moving on.\n     */\n    async release(): Promise<void> {\n        await this._session.release().catch(() => undefined);\n    }\n\n    /** The underlying `onnxruntime-web` session, for advanced use cases. */\n    get raw(): ort.InferenceSession {\n        return this._session;\n    }\n\n    /**\n     * Run inference and return all outputs.\n     *\n     * @param feeds Map of input name to `ort.Tensor`. Keys must match {@link inputNames}.\n     * @throws {@link InferenceError} if ORT raises any error during execution.\n     */\n    async run(feeds: Record<string, ort.Tensor>): Promise<Record<string, ort.Tensor>> {\n        try {\n            const result = await this._session.run(feeds);\n            return result as Record<string, ort.Tensor>;\n        } catch (err) {\n            throw new InferenceError(`Inference failed: ${(err as Error).message}`, { cause: err });\n        }\n    }\n}\n"],"mappings":";;;;;AA4BA,IAAa,IAAb,MAAa,EAAW;CAEC;CACD;CAFpB,YACI,GACA,GACF;EADkB,AADC,KAAA,WAAA,GACD,KAAA,YAAA;CACjB;CASH,aAAa,OAAO,GAAoB,IAA6B,CAAC,GAAwB;EAC1F,IAAM,IAAY,EAAiB,EAAQ,SAAS,GAC9C,IAAsD;GACxD,GAAI,EAAQ,kBAAkB,CAAC;GAC/B,oBACI;EACR,GAEI;EACJ,IAAI;GACA,AAKI,KALA,OAAO,KAAU,YAEV,aAAiB,YADd,MAAM,EAAW,iBAAiB,OAAO,GAAO,CAAc;EAShF,SAAS,GAAK;GACV,MAAM,IAAI,EAAe,8BAA+B,EAAc,WAAW,EAC7E,OAAO,EACX,CAAC;EACL;EAEA,OAAO,IAAI,EAAW,GAAS,CAAS;CAC5C;CAGA,IAAI,aAAgC;EAChC,OAAO,KAAK,SAAS;CACzB;CAGA,IAAI,YAAoB;EACpB,IAAM,IAAO,KAAK,SAAS,WAAW;EACtC,IAAI,MAAS,KAAA,GACT,MAAM,IAAI,EAAe,sBAAsB;EAEnD,OAAO;CACX;CAGA,IAAI,cAAiC;EACjC,OAAO,KAAK,SAAS;CACzB;CASA,IAAI,cAAwC;EACxC,OAAO,EACH,KAAK,SAAS,aAElB;CACJ;CAOA,IAAI,aAA4B;EAC5B,OAAO,KAAK,YAAY,MAAM,CAAC;CACnC;CAUA,MAAM,UAAyB;EAC3B,MAAM,KAAK,SAAS,QAAQ,CAAC,CAAC,YAAY,KAAA,CAAS;CACvD;CAGA,IAAI,MAA4B;EAC5B,OAAO,KAAK;CAChB;CAQA,MAAM,IAAI,GAAwE;EAC9E,IAAI;GAEA,OAAO,MADc,KAAK,SAAS,IAAI,CAAK;EAEhD,SAAS,GAAK;GACV,MAAM,IAAI,EAAe,qBAAsB,EAAc,WAAW,EAAE,OAAO,EAAI,CAAC;EAC1F;CACJ;AACJ"}

package/dist/vision/index.cjs

@@ -1,2 +1,2 @@
-const e=require("./core/exceptions.cjs"),t=require("./types.cjs"),n=require("./core/timing.cjs"),r=require("./results.cjs"),i=require("./labels.cjs"),a=require("./core/providers.cjs"),o=require("./core/session.cjs"),s=require("./io/image.cjs"),c=require("./preprocess/image.cjs"),l=require("./postprocess/classification.cjs"),u=require("./postprocess/detection.cjs"),d=require("./postprocess/segmentation.cjs"),f=require("./tasks/base.cjs"),p=require("./tasks/classifier.cjs"),m=require("./tasks/detector.cjs"),h=require("./tasks/segmenter.cjs");var g=`0.3.0`;exports.BoundingBox=t.BoundingBox,exports.Boxes=r.Boxes,exports.COCO_CLASSES=i.COCO_CLASSES,exports.ClassificationResults=r.ClassificationResults,exports.Classifier=p.Classifier,exports.DEFAULT_PROVIDERS=a.DEFAULT_PROVIDERS,exports.DetectionResults=r.DetectionResults,exports.Detector=m.Detector,exports.ImageLoadError=e.ImageLoadError,exports.InferenceError=e.InferenceError,exports.LabelMapError=e.LabelMapError,exports.Mask=t.Mask,exports.Masks=r.Masks,exports.ModelLoadError=e.ModelLoadError,exports.OrtSession=o.OrtSession,exports.OrtVisionError=e.OrtVisionError,exports.Probs=r.Probs,exports.ProviderNotAvailableError=e.ProviderNotAvailableError,exports.RGBImage=t.RGBImage,exports.SegmentationResults=r.SegmentationResults,exports.Segmenter=h.Segmenter,exports.SpeedTimer=n.SpeedTimer,exports.VERSION=g,exports.VisionTask=f.VisionTask,exports.batchedNms=u.batchedNms,exports.decodeYolo=u.decodeYolo,exports.decodeYoloAnchors=u.decodeYoloAnchors,exports.decodeYoloSeg=d.decodeYoloSeg,exports.decodeYoloV8=u.decodeYoloV8,exports.decodeYoloV8Anchors=u.decodeYoloV8Anchors,exports.decodeYoloV8Seg=d.decodeYoloV8Seg,exports.fromCv2=c.fromCv2,exports.letterbox=c.letterbox,exports.loadImage=s.loadImage,exports.nms=u.nms,exports.normalize=c.normalize,exports.resize=c.resize,exports.resolveLabels=i.resolveLabels,exports.resolveProviders=a.resolveProviders,exports.softmax=l.softmax,exports.toCHW=c.toCHW,exports.toCv2=c.toCv2,exports.toFloat32=c.toFloat32,exports.toFloat32Tensor=c.toFloat32Tensor,exports.toTensor=c.toTensor,exports.topK=l.topK;
+const e=require("./core/exceptions.cjs"),t=require("./types.cjs"),n=require("./core/timing.cjs"),r=require("./results.cjs"),i=require("./labels.cjs"),a=require("./core/graph.cjs"),o=require("./core/providers.cjs"),s=require("./core/session.cjs"),c=require("./io/image.cjs"),l=require("./preprocess/image.cjs"),u=require("./postprocess/classification.cjs"),d=require("./postprocess/detection.cjs"),f=require("./postprocess/segmentation.cjs"),p=require("./tasks/base.cjs"),m=require("./tasks/classifier.cjs"),h=require("./tasks/detector.cjs"),g=require("./tasks/segmenter.cjs");var _=`0.4.0`;exports.BoundingBox=t.BoundingBox,exports.Boxes=r.Boxes,exports.COCO_CLASSES=i.COCO_CLASSES,exports.ClassificationResults=r.ClassificationResults,exports.Classifier=m.Classifier,exports.DEFAULT_PROVIDERS=o.DEFAULT_PROVIDERS,exports.DetectionResults=r.DetectionResults,exports.Detector=h.Detector,exports.ImageLoadError=e.ImageLoadError,exports.InferenceError=e.InferenceError,exports.LabelMapError=e.LabelMapError,exports.Mask=t.Mask,exports.Masks=r.Masks,exports.ModelLoadError=e.ModelLoadError,exports.OrtSession=s.OrtSession,exports.OrtVisionError=e.OrtVisionError,exports.Probs=r.Probs,exports.ProviderNotAvailableError=e.ProviderNotAvailableError,exports.RGBImage=t.RGBImage,exports.SegmentationResults=r.SegmentationResults,exports.Segmenter=g.Segmenter,exports.SpeedTimer=n.SpeedTimer,exports.VERSION=_,exports.VisionTask=p.VisionTask,exports.batchedNms=d.batchedNms,exports.declaredShapesFrom=a.declaredShapesFrom,exports.decodeYolo=d.decodeYolo,exports.decodeYoloAnchors=d.decodeYoloAnchors,exports.decodeYoloSeg=f.decodeYoloSeg,exports.decodeYoloV8=d.decodeYoloV8,exports.decodeYoloV8Anchors=d.decodeYoloV8Anchors,exports.decodeYoloV8Seg=f.decodeYoloV8Seg,exports.fromCv2=l.fromCv2,exports.letterbox=l.letterbox,exports.loadImage=c.loadImage,exports.nms=d.nms,exports.normalize=l.normalize,exports.resize=l.resize,exports.resolveInputSize=a.resolveInputSize,exports.resolveLabels=i.resolveLabels,exports.resolveProviders=o.resolveProviders,exports.softmax=u.softmax,exports.spatialInputSize=a.spatialInputSize,exports.toCHW=l.toCHW,exports.toCv2=l.toCv2,exports.toFloat32=l.toFloat32,exports.toFloat32Tensor=l.toFloat32Tensor,exports.toTensor=l.toTensor,exports.topK=u.topK;
 //# sourceMappingURL=index.cjs.map

package/dist/vision/index.cjs.map

@@ -1 +1 @@
-{"version":3,"file":"index.cjs","names":[],"sources":["../../src/vision/index.ts"],"sourcesContent":["/**\n * `tempest-react-sdk/vision` — browser computer-vision inference with ONNX\n * Runtime Web (classification, detection, segmentation).\n *\n * Vendored from `@mauriciobenjamin700/ort-vision-sdk-web@0.3.0` (MIT, same\n * author) so it ships inside this SDK without an extra package install.\n * `onnxruntime-web` stays an optional peer dependency — install it (and ship\n * the matching `.wasm` files) only when you use this subpath.\n *\n * Do not hand-edit — regenerate with `npm run vendor:vision`.\n */\n\nexport {\n    BoundingBox,\n    Mask,\n    RGBImage,\n    type ClassProbability,\n    type ClassificationResult,\n    type DetectionResult,\n    type SegmentationResult,\n} from \"./types\";\n\nexport {\n    Boxes,\n    ClassificationResults,\n    DetectionResults,\n    Masks,\n    Probs,\n    SegmentationResults,\n} from \"./results\";\n\nexport { COCO_CLASSES, type LabelSpec, type ResolveLabelsOptions, resolveLabels } from \"./labels\";\n\nexport {\n    ImageLoadError,\n    InferenceError,\n    LabelMapError,\n    ModelLoadError,\n    OrtVisionError,\n    ProviderNotAvailableError,\n} from \"./core/exceptions\";\n\nexport { type ModelSource, type OrtSessionOptions, OrtSession } from \"./core/session\";\nexport { DEFAULT_PROVIDERS, resolveProviders } from \"./core/providers\";\nexport { type Speed, SpeedTimer } from \"./core/timing\";\n\nexport { type ImageInput, loadImage } from \"./io/image\";\n\nexport {\n    type LetterboxResult,\n    fromCv2,\n    letterbox,\n    normalize,\n    resize,\n    toCHW,\n    toCv2,\n    toFloat32,\n    toFloat32Tensor,\n    toTensor,\n} from \"./preprocess/image\";\n\nexport { type TopKResult, softmax, topK } from \"./postprocess/classification\";\n\nexport {\n    type DecodeYoloAnchorsOptions,\n    type DecodeYoloOptions,\n    type DecodeYoloV8AnchorsOptions,\n    type DecodeYoloV8Options,\n    type DecodedAnchors,\n    type DecodedDetection,\n    batchedNms,\n    decodeYolo,\n    decodeYoloAnchors,\n    decodeYoloV8,\n    decodeYoloV8Anchors,\n    nms,\n} from \"./postprocess/detection\";\n\nexport {\n    type DecodeYoloSegOptions,\n    type DecodeYoloV8SegOptions,\n    type DecodedSegmentation,\n    decodeYoloSeg,\n    decodeYoloV8Seg,\n} from \"./postprocess/segmentation\";\n\nexport { VisionTask } from \"./tasks/base\";\nexport {\n    type ClassifierOptions,\n    type ClassifierPredictOptions,\n    Classifier,\n} from \"./tasks/classifier\";\nexport {\n    type DetectorHead,\n    type DetectorOptions,\n    type DetectorPredictOptions,\n    Detector,\n} from \"./tasks/detector\";\nexport {\n    type SegmenterHead,\n    type SegmenterOptions,\n    type SegmenterPredictOptions,\n    Segmenter,\n} from \"./tasks/segmenter\";\n\nexport const VERSION: string = \"0.3.0\";\n"],"mappings":"kiBAyGA,IAAa,EAAkB"}
+{"version":3,"file":"index.cjs","names":[],"sources":["../../src/vision/index.ts"],"sourcesContent":["/**\n * `tempest-react-sdk/vision` — browser computer-vision inference with ONNX\n * Runtime Web (classification, detection, segmentation).\n *\n * Vendored from `@mauriciobenjamin700/ort-vision-sdk-web@0.4.0` (MIT, same\n * author) so it ships inside this SDK without an extra package install.\n * `onnxruntime-web` stays an optional peer dependency — install it (and ship\n * the matching `.wasm` files) only when you use this subpath.\n *\n * Do not hand-edit — regenerate with `npm run vendor:vision`.\n */\n\nexport {\n    BoundingBox,\n    Mask,\n    RGBImage,\n    type ClassProbability,\n    type ClassificationResult,\n    type DetectionResult,\n    type SegmentationResult,\n} from \"./types\";\n\nexport {\n    Boxes,\n    ClassificationResults,\n    DetectionResults,\n    Masks,\n    Probs,\n    SegmentationResults,\n} from \"./results\";\n\nexport { COCO_CLASSES, type LabelSpec, type ResolveLabelsOptions, resolveLabels } from \"./labels\";\n\nexport {\n    ImageLoadError,\n    InferenceError,\n    LabelMapError,\n    ModelLoadError,\n    OrtVisionError,\n    ProviderNotAvailableError,\n} from \"./core/exceptions\";\n\nexport { type ModelSource, type OrtSessionOptions, OrtSession } from \"./core/session\";\nexport {\n    type DeclaredDim,\n    type DeclaredShape,\n    type ResolveInputSizeOptions,\n    declaredShapesFrom,\n    resolveInputSize,\n    spatialInputSize,\n} from \"./core/graph\";\nexport { DEFAULT_PROVIDERS, resolveProviders } from \"./core/providers\";\nexport { type Speed, SpeedTimer } from \"./core/timing\";\n\nexport { type ImageInput, loadImage } from \"./io/image\";\n\nexport {\n    type LetterboxResult,\n    fromCv2,\n    letterbox,\n    normalize,\n    resize,\n    toCHW,\n    toCv2,\n    toFloat32,\n    toFloat32Tensor,\n    toTensor,\n} from \"./preprocess/image\";\n\nexport { type TopKResult, softmax, topK } from \"./postprocess/classification\";\n\nexport {\n    type DecodeYoloAnchorsOptions,\n    type DecodeYoloOptions,\n    type DecodeYoloV8AnchorsOptions,\n    type DecodeYoloV8Options,\n    type DecodedAnchors,\n    type DecodedDetection,\n    batchedNms,\n    decodeYolo,\n    decodeYoloAnchors,\n    decodeYoloV8,\n    decodeYoloV8Anchors,\n    nms,\n} from \"./postprocess/detection\";\n\nexport {\n    type DecodeYoloSegOptions,\n    type DecodeYoloV8SegOptions,\n    type DecodedSegmentation,\n    decodeYoloSeg,\n    decodeYoloV8Seg,\n} from \"./postprocess/segmentation\";\n\nexport { VisionTask } from \"./tasks/base\";\nexport {\n    type ClassifierOptions,\n    type ClassifierPredictOptions,\n    Classifier,\n} from \"./tasks/classifier\";\nexport {\n    type DetectorHead,\n    type DetectorOptions,\n    type DetectorPredictOptions,\n    Detector,\n} from \"./tasks/detector\";\nexport {\n    type SegmenterHead,\n    type SegmenterOptions,\n    type SegmenterPredictOptions,\n    Segmenter,\n} from \"./tasks/segmenter\";\n\nexport const VERSION: string = \"0.4.0\";\n"],"mappings":"gkBAiHA,IAAa,EAAkB"}

package/dist/vision/index.js

@@ -3,20 +3,21 @@ import { BoundingBox as o, Mask as s, RGBImage as c } from "./types.js";
 import { SpeedTimer as l } from "./core/timing.js";
 import { Boxes as u, ClassificationResults as d, DetectionResults as f, Masks as p, Probs as m, SegmentationResults as h } from "./results.js";
 import { COCO_CLASSES as g, resolveLabels as _ } from "./labels.js";
-import { DEFAULT_PROVIDERS as v, resolveProviders as y } from "./core/providers.js";
-import { OrtSession as b } from "./core/session.js";
-import { loadImage as x } from "./io/image.js";
-import { fromCv2 as S, letterbox as C, normalize as w, resize as T, toCHW as E, toCv2 as D, toFloat32 as O, toFloat32Tensor as k, toTensor as A } from "./preprocess/image.js";
-import { softmax as j, topK as M } from "./postprocess/classification.js";
-import { batchedNms as N, decodeYolo as P, decodeYoloAnchors as F, decodeYoloV8 as I, decodeYoloV8Anchors as L, nms as R } from "./postprocess/detection.js";
-import { decodeYoloSeg as z, decodeYoloV8Seg as B } from "./postprocess/segmentation.js";
-import { VisionTask as V } from "./tasks/base.js";
-import { Classifier as H } from "./tasks/classifier.js";
-import { Detector as U } from "./tasks/detector.js";
-import { Segmenter as W } from "./tasks/segmenter.js";
+import { declaredShapesFrom as v, resolveInputSize as y, spatialInputSize as b } from "./core/graph.js";
+import { DEFAULT_PROVIDERS as x, resolveProviders as S } from "./core/providers.js";
+import { OrtSession as C } from "./core/session.js";
+import { loadImage as w } from "./io/image.js";
+import { fromCv2 as T, letterbox as E, normalize as D, resize as O, toCHW as k, toCv2 as A, toFloat32 as j, toFloat32Tensor as M, toTensor as N } from "./preprocess/image.js";
+import { softmax as P, topK as F } from "./postprocess/classification.js";
+import { batchedNms as I, decodeYolo as L, decodeYoloAnchors as R, decodeYoloV8 as z, decodeYoloV8Anchors as B, nms as V } from "./postprocess/detection.js";
+import { decodeYoloSeg as H, decodeYoloV8Seg as U } from "./postprocess/segmentation.js";
+import { VisionTask as W } from "./tasks/base.js";
+import { Classifier as G } from "./tasks/classifier.js";
+import { Detector as K } from "./tasks/detector.js";
+import { Segmenter as q } from "./tasks/segmenter.js";
 //#region src/vision/index.ts
-var G = "0.3.0";
+var J = "0.4.0";
 //#endregion
-export { o as BoundingBox, u as Boxes, g as COCO_CLASSES, d as ClassificationResults, H as Classifier, v as DEFAULT_PROVIDERS, f as DetectionResults, U as Detector, e as ImageLoadError, t as InferenceError, n as LabelMapError, s as Mask, p as Masks, r as ModelLoadError, b as OrtSession, i as OrtVisionError, m as Probs, a as ProviderNotAvailableError, c as RGBImage, h as SegmentationResults, W as Segmenter, l as SpeedTimer, G as VERSION, V as VisionTask, N as batchedNms, P as decodeYolo, F as decodeYoloAnchors, z as decodeYoloSeg, I as decodeYoloV8, L as decodeYoloV8Anchors, B as decodeYoloV8Seg, S as fromCv2, C as letterbox, x as loadImage, R as nms, w as normalize, T as resize, _ as resolveLabels, y as resolveProviders, j as softmax, E as toCHW, D as toCv2, O as toFloat32, k as toFloat32Tensor, A as toTensor, M as topK };
+export { o as BoundingBox, u as Boxes, g as COCO_CLASSES, d as ClassificationResults, G as Classifier, x as DEFAULT_PROVIDERS, f as DetectionResults, K as Detector, e as ImageLoadError, t as InferenceError, n as LabelMapError, s as Mask, p as Masks, r as ModelLoadError, C as OrtSession, i as OrtVisionError, m as Probs, a as ProviderNotAvailableError, c as RGBImage, h as SegmentationResults, q as Segmenter, l as SpeedTimer, J as VERSION, W as VisionTask, I as batchedNms, v as declaredShapesFrom, L as decodeYolo, R as decodeYoloAnchors, H as decodeYoloSeg, z as decodeYoloV8, B as decodeYoloV8Anchors, U as decodeYoloV8Seg, T as fromCv2, E as letterbox, w as loadImage, V as nms, D as normalize, O as resize, y as resolveInputSize, _ as resolveLabels, S as resolveProviders, P as softmax, b as spatialInputSize, k as toCHW, A as toCv2, j as toFloat32, M as toFloat32Tensor, N as toTensor, F as topK };
 
 //# sourceMappingURL=index.js.map

package/dist/vision/index.js.map

@@ -1 +1 @@
-{"version":3,"file":"index.js","names":[],"sources":["../../src/vision/index.ts"],"sourcesContent":["/**\n * `tempest-react-sdk/vision` — browser computer-vision inference with ONNX\n * Runtime Web (classification, detection, segmentation).\n *\n * Vendored from `@mauriciobenjamin700/ort-vision-sdk-web@0.3.0` (MIT, same\n * author) so it ships inside this SDK without an extra package install.\n * `onnxruntime-web` stays an optional peer dependency — install it (and ship\n * the matching `.wasm` files) only when you use this subpath.\n *\n * Do not hand-edit — regenerate with `npm run vendor:vision`.\n */\n\nexport {\n    BoundingBox,\n    Mask,\n    RGBImage,\n    type ClassProbability,\n    type ClassificationResult,\n    type DetectionResult,\n    type SegmentationResult,\n} from \"./types\";\n\nexport {\n    Boxes,\n    ClassificationResults,\n    DetectionResults,\n    Masks,\n    Probs,\n    SegmentationResults,\n} from \"./results\";\n\nexport { COCO_CLASSES, type LabelSpec, type ResolveLabelsOptions, resolveLabels } from \"./labels\";\n\nexport {\n    ImageLoadError,\n    InferenceError,\n    LabelMapError,\n    ModelLoadError,\n    OrtVisionError,\n    ProviderNotAvailableError,\n} from \"./core/exceptions\";\n\nexport { type ModelSource, type OrtSessionOptions, OrtSession } from \"./core/session\";\nexport { DEFAULT_PROVIDERS, resolveProviders } from \"./core/providers\";\nexport { type Speed, SpeedTimer } from \"./core/timing\";\n\nexport { type ImageInput, loadImage } from \"./io/image\";\n\nexport {\n    type LetterboxResult,\n    fromCv2,\n    letterbox,\n    normalize,\n    resize,\n    toCHW,\n    toCv2,\n    toFloat32,\n    toFloat32Tensor,\n    toTensor,\n} from \"./preprocess/image\";\n\nexport { type TopKResult, softmax, topK } from \"./postprocess/classification\";\n\nexport {\n    type DecodeYoloAnchorsOptions,\n    type DecodeYoloOptions,\n    type DecodeYoloV8AnchorsOptions,\n    type DecodeYoloV8Options,\n    type DecodedAnchors,\n    type DecodedDetection,\n    batchedNms,\n    decodeYolo,\n    decodeYoloAnchors,\n    decodeYoloV8,\n    decodeYoloV8Anchors,\n    nms,\n} from \"./postprocess/detection\";\n\nexport {\n    type DecodeYoloSegOptions,\n    type DecodeYoloV8SegOptions,\n    type DecodedSegmentation,\n    decodeYoloSeg,\n    decodeYoloV8Seg,\n} from \"./postprocess/segmentation\";\n\nexport { VisionTask } from \"./tasks/base\";\nexport {\n    type ClassifierOptions,\n    type ClassifierPredictOptions,\n    Classifier,\n} from \"./tasks/classifier\";\nexport {\n    type DetectorHead,\n    type DetectorOptions,\n    type DetectorPredictOptions,\n    Detector,\n} from \"./tasks/detector\";\nexport {\n    type SegmenterHead,\n    type SegmenterOptions,\n    type SegmenterPredictOptions,\n    Segmenter,\n} from \"./tasks/segmenter\";\n\nexport const VERSION: string = \"0.3.0\";\n"],"mappings":";;;;;;;;;;;;;;;;;AAyGA,IAAa,IAAkB"}
+{"version":3,"file":"index.js","names":[],"sources":["../../src/vision/index.ts"],"sourcesContent":["/**\n * `tempest-react-sdk/vision` — browser computer-vision inference with ONNX\n * Runtime Web (classification, detection, segmentation).\n *\n * Vendored from `@mauriciobenjamin700/ort-vision-sdk-web@0.4.0` (MIT, same\n * author) so it ships inside this SDK without an extra package install.\n * `onnxruntime-web` stays an optional peer dependency — install it (and ship\n * the matching `.wasm` files) only when you use this subpath.\n *\n * Do not hand-edit — regenerate with `npm run vendor:vision`.\n */\n\nexport {\n    BoundingBox,\n    Mask,\n    RGBImage,\n    type ClassProbability,\n    type ClassificationResult,\n    type DetectionResult,\n    type SegmentationResult,\n} from \"./types\";\n\nexport {\n    Boxes,\n    ClassificationResults,\n    DetectionResults,\n    Masks,\n    Probs,\n    SegmentationResults,\n} from \"./results\";\n\nexport { COCO_CLASSES, type LabelSpec, type ResolveLabelsOptions, resolveLabels } from \"./labels\";\n\nexport {\n    ImageLoadError,\n    InferenceError,\n    LabelMapError,\n    ModelLoadError,\n    OrtVisionError,\n    ProviderNotAvailableError,\n} from \"./core/exceptions\";\n\nexport { type ModelSource, type OrtSessionOptions, OrtSession } from \"./core/session\";\nexport {\n    type DeclaredDim,\n    type DeclaredShape,\n    type ResolveInputSizeOptions,\n    declaredShapesFrom,\n    resolveInputSize,\n    spatialInputSize,\n} from \"./core/graph\";\nexport { DEFAULT_PROVIDERS, resolveProviders } from \"./core/providers\";\nexport { type Speed, SpeedTimer } from \"./core/timing\";\n\nexport { type ImageInput, loadImage } from \"./io/image\";\n\nexport {\n    type LetterboxResult,\n    fromCv2,\n    letterbox,\n    normalize,\n    resize,\n    toCHW,\n    toCv2,\n    toFloat32,\n    toFloat32Tensor,\n    toTensor,\n} from \"./preprocess/image\";\n\nexport { type TopKResult, softmax, topK } from \"./postprocess/classification\";\n\nexport {\n    type DecodeYoloAnchorsOptions,\n    type DecodeYoloOptions,\n    type DecodeYoloV8AnchorsOptions,\n    type DecodeYoloV8Options,\n    type DecodedAnchors,\n    type DecodedDetection,\n    batchedNms,\n    decodeYolo,\n    decodeYoloAnchors,\n    decodeYoloV8,\n    decodeYoloV8Anchors,\n    nms,\n} from \"./postprocess/detection\";\n\nexport {\n    type DecodeYoloSegOptions,\n    type DecodeYoloV8SegOptions,\n    type DecodedSegmentation,\n    decodeYoloSeg,\n    decodeYoloV8Seg,\n} from \"./postprocess/segmentation\";\n\nexport { VisionTask } from \"./tasks/base\";\nexport {\n    type ClassifierOptions,\n    type ClassifierPredictOptions,\n    Classifier,\n} from \"./tasks/classifier\";\nexport {\n    type DetectorHead,\n    type DetectorOptions,\n    type DetectorPredictOptions,\n    Detector,\n} from \"./tasks/detector\";\nexport {\n    type SegmenterHead,\n    type SegmenterOptions,\n    type SegmenterPredictOptions,\n    Segmenter,\n} from \"./tasks/segmenter\";\n\nexport const VERSION: string = \"0.4.0\";\n"],"mappings":";;;;;;;;;;;;;;;;;;AAiHA,IAAa,IAAkB"}

package/dist/vision/tasks/classifier.cjs

@@ -1,2 +1,2 @@
-const e=require("../core/timing.cjs"),t=require("../results.cjs"),n=require("../labels.cjs"),r=require("../core/session.cjs"),i=require("../io/image.cjs"),a=require("../preprocess/image.cjs"),o=require("../postprocess/classification.cjs"),s=require("./base.cjs");var c=[.485,.456,.406],l=[.229,.224,.225],u=class u extends s.VisionTask{_labels;_names;_inputSize;_mean;_std;_applySoftmax;constructor(e,t,n,r,i,a,o){super(e),this._labels=t,this._names=n,this._inputSize=r,this._mean=i,this._std=a,this._applySoftmax=o}static async create(e,t){let i=await r.OrtSession.create(e,t),a=n.resolveLabels(t.labels,{numClasses:t.numClasses}),o={};for(let e=0;e<a.length;e++)o[e]=a[e];return new u(i,a,o,t.inputSize??[224,224],t.mean??c,t.std??l,t.applySoftmax??!0)}get labels(){return this._labels}get names(){return this._names}get numClasses(){return this._labels.length}async call(e,t={}){return this.predict(e,t)}async predict(n,r={}){let a=new e.SpeedTimer,s=typeof n==`string`?n:null,c=await i.loadImage(n);a.stage(`load`);let l=this._preprocess(c);a.stage(`preprocess`);let u=await this._session.run({[this._session.inputName]:l});a.stage(`inference`);let d=this._session.outputNames[0];if(d===void 0)throw Error(`Classifier model has no outputs.`);let f=u[d];if(f===void 0)throw Error(`Classifier model output ${d} missing from run() result.`);let p=this._postprocess(f.data),{indices:m,values:h}=o.topK(p,r.topK??null),g=[];for(let e=0;e<m.length;e++){let t=m[e],n=this._labels[t]??`class_${t}`;g.push({classId:t,className:n,probability:h[e],cls:t,name:n,conf:h[e]})}if(g.length===0)throw Error(`Classifier produced no probabilities (empty output).`);let _=g[0],v={classId:_.classId,className:_.className,confidence:_.probability,cls:_.classId,name:_.className,conf:_.probability,image:c,probabilities:g},y=[c.height,c.width],b=new t.Probs(p);return a.stage(`postprocess`),[new t.ClassificationResults(b,v,this._names,c,y,s,a.speed())]}_preprocess(e){let[t,n]=this._inputSize,r=a.resize(e,t,n);return a.toFloat32Tensor(a.toCHW(a.normalize(r,this._mean,this._std),r.width,r.height,3),[1,3,r.height,r.width])}_postprocess(e){return this._applySoftmax?o.softmax(e):new Float32Array(e)}};exports.Classifier=u;
+const e=require("../core/timing.cjs"),t=require("../results.cjs"),n=require("../labels.cjs"),r=require("../core/graph.cjs"),i=require("../core/session.cjs"),a=require("../io/image.cjs"),o=require("../preprocess/image.cjs"),s=require("../postprocess/classification.cjs"),c=require("./base.cjs");var l=[.485,.456,.406],u=[.229,.224,.225],d=class d extends c.VisionTask{_labels;_names;_inputSize;_mean;_std;_applySoftmax;constructor(e,t,n,r,i,a,o){super(e),this._labels=t,this._names=n,this._inputSize=r,this._mean=i,this._std=a,this._applySoftmax=o}static async create(e,t){let a=await i.OrtSession.create(e,t),o=n.resolveLabels(t.labels,{numClasses:t.numClasses}),s={};for(let e=0;e<o.length;e++)s[e]=o[e];return new d(a,o,s,r.resolveInputSize({graphShape:a.inputShape,requested:t.inputSize,fallback:[224,224]}),t.mean??l,t.std??u,t.applySoftmax??!0)}get labels(){return this._labels}get names(){return this._names}get inputSize(){return this._inputSize}get numClasses(){return this._labels.length}async call(e,t={}){return this.predict(e,t)}async predict(n,r={}){let i=new e.SpeedTimer,o=typeof n==`string`?n:null,c=await a.loadImage(n);i.stage(`load`);let l=this._preprocess(c);i.stage(`preprocess`);let u=await this._session.run({[this._session.inputName]:l});i.stage(`inference`);let d=this._session.outputNames[0];if(d===void 0)throw Error(`Classifier model has no outputs.`);let f=u[d];if(f===void 0)throw Error(`Classifier model output ${d} missing from run() result.`);let p=this._postprocess(f.data),{indices:m,values:h}=s.topK(p,r.topK??null),g=[];for(let e=0;e<m.length;e++){let t=m[e],n=this._labels[t]??`class_${t}`;g.push({classId:t,className:n,probability:h[e],cls:t,name:n,conf:h[e]})}if(g.length===0)throw Error(`Classifier produced no probabilities (empty output).`);let _=g[0],v={classId:_.classId,className:_.className,confidence:_.probability,cls:_.classId,name:_.className,conf:_.probability,image:c,probabilities:g},y=[c.height,c.width],b=new t.Probs(p);return i.stage(`postprocess`),[new t.ClassificationResults(b,v,this._names,c,y,o,i.speed())]}_preprocess(e){let[t,n]=this._inputSize,r=o.resize(e,t,n);return o.toFloat32Tensor(o.toCHW(o.normalize(r,this._mean,this._std),r.width,r.height,3),[1,3,r.height,r.width])}_postprocess(e){return this._applySoftmax?s.softmax(e):new Float32Array(e)}};exports.Classifier=d;
 //# sourceMappingURL=classifier.cjs.map

package/dist/vision/tasks/classifier.cjs.map

@@ -1 +1 @@
-{"version":3,"file":"classifier.cjs","names":[],"sources":["../../../src/vision/tasks/classifier.ts"],"sourcesContent":["/**\n * Image classification task using ONNX Runtime Web.\n */\n\nimport type * as ort from \"onnxruntime-web\";\n\nimport { type ModelSource, type OrtSessionOptions, OrtSession } from \"../core/session\";\nimport { SpeedTimer } from \"../core/timing\";\nimport { type ImageInput, loadImage } from \"../io/image\";\nimport { type LabelSpec, resolveLabels } from \"../labels\";\nimport { softmax, topK } from \"../postprocess/classification\";\nimport { normalize, resize, toCHW, toFloat32Tensor } from \"../preprocess/image\";\nimport { ClassificationResults, Probs } from \"../results\";\nimport { VisionTask } from \"./base\";\nimport { type ClassProbability, type ClassificationResult, type RGBImage } from \"../types\";\n\nconst IMAGENET_MEAN: readonly [number, number, number] = [0.485, 0.456, 0.406];\nconst IMAGENET_STD: readonly [number, number, number] = [0.229, 0.224, 0.225];\n\nexport interface ClassifierOptions extends OrtSessionOptions {\n    /** Class label spec — see {@link resolveLabels}. */\n    readonly labels: LabelSpec;\n    /**\n     * Number of classes the model can predict. Required when `labels` is `null`\n     * or when you want to validate that the supplied labels match the model.\n     */\n    readonly numClasses?: number;\n    /** Model input `[width, height]` in pixels. Defaults to `[224, 224]`. */\n    readonly inputSize?: readonly [number, number];\n    /** Per-channel RGB mean used for normalization. Defaults to ImageNet. */\n    readonly mean?: readonly [number, number, number];\n    /** Per-channel RGB standard deviation. Defaults to ImageNet. */\n    readonly std?: readonly [number, number, number];\n    /**\n     * If `true` (default), apply softmax to the raw model output. Set to\n     * `false` for models whose final layer already produces a probability\n     * distribution.\n     */\n    readonly applySoftmax?: boolean;\n}\n\nexport interface ClassifierPredictOptions {\n    /**\n     * If set, the per-class probability list in `results[0].result.probabilities`\n     * is truncated to the top-K entries. The bulk `probs` view always exposes\n     * the full vector.\n     */\n    readonly topK?: number;\n}\n\n/**\n * Image classifier wrapping an ONNX model with ImageNet-style preprocessing.\n *\n * `predict()` returns `Promise<ClassificationResults[]>` (length 1 for a\n * single image), mirroring Ultralytics' API. The envelope exposes a `probs`\n * collection (`top1`, `top1conf`, `top5`, `top5conf`, `data`) plus the\n * legacy per-class probability list with names already resolved.\n *\n * Defaults: 224×224 RGB input, `float32` normalized with ImageNet mean/std,\n * NCHW layout, batch size 1, softmax applied to the raw output.\n *\n * @example\n * ```typescript\n * const clf = await Classifier.create(\"/models/resnet50.onnx\", {\n *   labels: [\"tench\", \"goldfish\", ...]  // 1000 ImageNet labels\n * });\n * const r = (await clf.predict(\"/images/dog.jpg\"))[0];\n * console.log(r.cls, r.conf, r.name);\n * console.log(r.probs.top5, r.probs.top5conf);\n * ```\n */\nexport class Classifier extends VisionTask {\n    private constructor(\n        session: OrtSession,\n        private readonly _labels: readonly string[],\n        private readonly _names: Readonly<Record<number, string>>,\n        private readonly _inputSize: readonly [number, number],\n        private readonly _mean: readonly [number, number, number],\n        private readonly _std: readonly [number, number, number],\n        private readonly _applySoftmax: boolean,\n    ) {\n        super(session);\n    }\n\n    /** Load the model and resolve labels. */\n    static async create(model: ModelSource, options: ClassifierOptions): Promise<Classifier> {\n        const session = await OrtSession.create(model, options);\n        const labels = resolveLabels(options.labels, { numClasses: options.numClasses });\n        const names: Record<number, string> = {};\n        for (let i = 0; i < labels.length; i++) {\n            names[i] = labels[i] as string;\n        }\n        return new Classifier(\n            session,\n            labels,\n            names,\n            options.inputSize ?? [224, 224],\n            options.mean ?? IMAGENET_MEAN,\n            options.std ?? IMAGENET_STD,\n            options.applySoftmax ?? true,\n        );\n    }\n\n    /** Class labels indexed by class id. */\n    get labels(): readonly string[] {\n        return this._labels;\n    }\n\n    /** Class id → class name dict (matches Ultralytics' `model.names`). */\n    get names(): Readonly<Record<number, string>> {\n        return this._names;\n    }\n\n    /** Number of classes the model can predict. */\n    get numClasses(): number {\n        return this._labels.length;\n    }\n\n    /** Alias for {@link predict} (parity with PyTorch `nn.Module.__call__`). */\n    async call(\n        image: ImageInput,\n        options: ClassifierPredictOptions = {},\n    ): Promise<ClassificationResults[]> {\n        return this.predict(image, options);\n    }\n\n    /** Run classification on a single image. */\n    async predict(\n        image: ImageInput,\n        options: ClassifierPredictOptions = {},\n    ): Promise<ClassificationResults[]> {\n        const timer = new SpeedTimer();\n        const path = typeof image === \"string\" ? image : null;\n        const original = await loadImage(image);\n        timer.stage(\"load\");\n        const tensor = this._preprocess(original);\n        timer.stage(\"preprocess\");\n        const outputs = await this._session.run({ [this._session.inputName]: tensor });\n        timer.stage(\"inference\");\n        const firstOutputName = this._session.outputNames[0];\n        if (firstOutputName === undefined) {\n            throw new Error(\"Classifier model has no outputs.\");\n        }\n        const raw = outputs[firstOutputName];\n        if (raw === undefined) {\n            throw new Error(\n                `Classifier model output ${firstOutputName} missing from run() result.`,\n            );\n        }\n        const fullProbs = this._postprocess(raw.data as Float32Array);\n\n        const { indices, values } = topK(fullProbs, options.topK ?? null);\n        const probabilities: ClassProbability[] = [];\n        for (let i = 0; i < indices.length; i++) {\n            const id = indices[i] as number;\n            const className = this._labels[id] ?? `class_${id}`;\n            probabilities.push({\n                classId: id,\n                className,\n                probability: values[i] as number,\n                cls: id,\n                name: className,\n                conf: values[i] as number,\n            });\n        }\n        if (probabilities.length === 0) {\n            throw new Error(\"Classifier produced no probabilities (empty output).\");\n        }\n\n        const top = probabilities[0] as ClassProbability;\n        const result: ClassificationResult = {\n            classId: top.classId,\n            className: top.className,\n            confidence: top.probability,\n            cls: top.classId,\n            name: top.className,\n            conf: top.probability,\n            image: original,\n            probabilities,\n        };\n\n        const orig: readonly [number, number] = [original.height, original.width];\n        const probs = new Probs(fullProbs);\n        timer.stage(\"postprocess\");\n        return [\n            new ClassificationResults(\n                probs,\n                result,\n                this._names,\n                original,\n                orig,\n                path,\n                timer.speed(),\n            ),\n        ];\n    }\n\n    private _preprocess(image: RGBImage): ort.Tensor {\n        const [tw, th] = this._inputSize;\n        const resized = resize(image, tw, th);\n        const normalized = normalize(resized, this._mean, this._std);\n        const chw = toCHW(normalized, resized.width, resized.height, 3);\n        return toFloat32Tensor(chw, [1, 3, resized.height, resized.width]);\n    }\n\n    private _postprocess(raw: Float32Array): Float32Array {\n        return this._applySoftmax ? softmax(raw) : new Float32Array(raw);\n    }\n}\n"],"mappings":"uQAgBA,IAAM,EAAmD,CAAC,KAAO,KAAO,IAAK,EACvE,EAAkD,CAAC,KAAO,KAAO,IAAK,EAsD/D,EAAb,MAAa,UAAmB,EAAA,UAAW,CAGlB,QACA,OACA,WACA,MACA,KACA,cAPrB,YACI,EACA,EACA,EACA,EACA,EACA,EACA,EACF,CACE,MAAM,CAAO,EAPI,KAAA,QAAA,EACA,KAAA,OAAA,EACA,KAAA,WAAA,EACA,KAAA,MAAA,EACA,KAAA,KAAA,EACA,KAAA,cAAA,CAGrB,CAGA,aAAa,OAAO,EAAoB,EAAiD,CACrF,IAAM,EAAU,MAAM,EAAA,WAAW,OAAO,EAAO,CAAO,EAChD,EAAS,EAAA,cAAc,EAAQ,OAAQ,CAAE,WAAY,EAAQ,UAAW,CAAC,EACzE,EAAgC,CAAC,EACvC,IAAK,IAAI,EAAI,EAAG,EAAI,EAAO,OAAQ,IAC/B,EAAM,GAAK,EAAO,GAEtB,OAAO,IAAI,EACP,EACA,EACA,EACA,EAAQ,WAAa,CAAC,IAAK,GAAG,EAC9B,EAAQ,MAAQ,EAChB,EAAQ,KAAO,EACf,EAAQ,cAAgB,EAC5B,CACJ,CAGA,IAAI,QAA4B,CAC5B,OAAO,KAAK,OAChB,CAGA,IAAI,OAA0C,CAC1C,OAAO,KAAK,MAChB,CAGA,IAAI,YAAqB,CACrB,OAAO,KAAK,QAAQ,MACxB,CAGA,MAAM,KACF,EACA,EAAoC,CAAC,EACL,CAChC,OAAO,KAAK,QAAQ,EAAO,CAAO,CACtC,CAGA,MAAM,QACF,EACA,EAAoC,CAAC,EACL,CAChC,IAAM,EAAQ,IAAI,EAAA,WACZ,EAAO,OAAO,GAAU,SAAW,EAAQ,KAC3C,EAAW,MAAM,EAAA,UAAU,CAAK,EACtC,EAAM,MAAM,MAAM,EAClB,IAAM,EAAS,KAAK,YAAY,CAAQ,EACxC,EAAM,MAAM,YAAY,EACxB,IAAM,EAAU,MAAM,KAAK,SAAS,IAAI,EAAG,KAAK,SAAS,WAAY,CAAO,CAAC,EAC7E,EAAM,MAAM,WAAW,EACvB,IAAM,EAAkB,KAAK,SAAS,YAAY,GAClD,GAAI,IAAoB,IAAA,GACpB,MAAU,MAAM,kCAAkC,EAEtD,IAAM,EAAM,EAAQ,GACpB,GAAI,IAAQ,IAAA,GACR,MAAU,MACN,2BAA2B,EAAgB,4BAC/C,EAEJ,IAAM,EAAY,KAAK,aAAa,EAAI,IAAoB,EAEtD,CAAE,UAAS,UAAW,EAAA,KAAK,EAAW,EAAQ,MAAQ,IAAI,EAC1D,EAAoC,CAAC,EAC3C,IAAK,IAAI,EAAI,EAAG,EAAI,EAAQ,OAAQ,IAAK,CACrC,IAAM,EAAK,EAAQ,GACb,EAAY,KAAK,QAAQ,IAAO,SAAS,IAC/C,EAAc,KAAK,CACf,QAAS,EACT,YACA,YAAa,EAAO,GACpB,IAAK,EACL,KAAM,EACN,KAAM,EAAO,EACjB,CAAC,CACL,CACA,GAAI,EAAc,SAAW,EACzB,MAAU,MAAM,sDAAsD,EAG1E,IAAM,EAAM,EAAc,GACpB,EAA+B,CACjC,QAAS,EAAI,QACb,UAAW,EAAI,UACf,WAAY,EAAI,YAChB,IAAK,EAAI,QACT,KAAM,EAAI,UACV,KAAM,EAAI,YACV,MAAO,EACP,eACJ,EAEM,EAAkC,CAAC,EAAS,OAAQ,EAAS,KAAK,EAClE,EAAQ,IAAI,EAAA,MAAM,CAAS,EAEjC,OADA,EAAM,MAAM,aAAa,EAClB,CACH,IAAI,EAAA,sBACA,EACA,EACA,KAAK,OACL,EACA,EACA,EACA,EAAM,MAAM,CAChB,CACJ,CACJ,CAEA,YAAoB,EAA6B,CAC7C,GAAM,CAAC,EAAI,GAAM,KAAK,WAChB,EAAU,EAAA,OAAO,EAAO,EAAI,CAAE,EAGpC,OAAO,EAAA,gBADK,EAAA,MADO,EAAA,UAAU,EAAS,KAAK,MAAO,KAAK,IACrC,EAAY,EAAQ,MAAO,EAAQ,OAAQ,CACtC,EAAK,CAAC,EAAG,EAAG,EAAQ,OAAQ,EAAQ,KAAK,CAAC,CACrE,CAEA,aAAqB,EAAiC,CAClD,OAAO,KAAK,cAAgB,EAAA,QAAQ,CAAG,EAAI,IAAI,aAAa,CAAG,CACnE,CACJ"}
+{"version":3,"file":"classifier.cjs","names":[],"sources":["../../../src/vision/tasks/classifier.ts"],"sourcesContent":["/**\n * Image classification task using ONNX Runtime Web.\n */\n\nimport type * as ort from \"onnxruntime-web\";\n\nimport { type ModelSource, type OrtSessionOptions, OrtSession } from \"../core/session\";\nimport { SpeedTimer } from \"../core/timing\";\nimport { resolveInputSize } from \"../core/graph\";\nimport { type ImageInput, loadImage } from \"../io/image\";\nimport { type LabelSpec, resolveLabels } from \"../labels\";\nimport { softmax, topK } from \"../postprocess/classification\";\nimport { normalize, resize, toCHW, toFloat32Tensor } from \"../preprocess/image\";\nimport { ClassificationResults, Probs } from \"../results\";\nimport { VisionTask } from \"./base\";\nimport { type ClassProbability, type ClassificationResult, type RGBImage } from \"../types\";\n\nconst IMAGENET_MEAN: readonly [number, number, number] = [0.485, 0.456, 0.406];\nconst IMAGENET_STD: readonly [number, number, number] = [0.229, 0.224, 0.225];\n\nexport interface ClassifierOptions extends OrtSessionOptions {\n    /** Class label spec — see {@link resolveLabels}. */\n    readonly labels: LabelSpec;\n    /**\n     * Number of classes the model can predict. Required when `labels` is `null`\n     * or when you want to validate that the supplied labels match the model.\n     */\n    readonly numClasses?: number;\n    /**\n     * Model input `[width, height]` in pixels.\n     *\n     * Only used when the model's graph leaves its spatial axes dynamic: a graph\n     * that declares a static size always wins, since that is the only shape ONNX\n     * Runtime will accept. Defaults to `[224, 224]`.\n     */\n    readonly inputSize?: readonly [number, number];\n    /** Per-channel RGB mean used for normalization. Defaults to ImageNet. */\n    readonly mean?: readonly [number, number, number];\n    /** Per-channel RGB standard deviation. Defaults to ImageNet. */\n    readonly std?: readonly [number, number, number];\n    /**\n     * If `true` (default), apply softmax to the raw model output. Set to\n     * `false` for models whose final layer already produces a probability\n     * distribution.\n     */\n    readonly applySoftmax?: boolean;\n}\n\nexport interface ClassifierPredictOptions {\n    /**\n     * If set, the per-class probability list in `results[0].result.probabilities`\n     * is truncated to the top-K entries. The bulk `probs` view always exposes\n     * the full vector.\n     */\n    readonly topK?: number;\n}\n\n/**\n * Image classifier wrapping an ONNX model with ImageNet-style preprocessing.\n *\n * `predict()` returns `Promise<ClassificationResults[]>` (length 1 for a\n * single image), mirroring Ultralytics' API. The envelope exposes a `probs`\n * collection (`top1`, `top1conf`, `top5`, `top5conf`, `data`) plus the\n * legacy per-class probability list with names already resolved.\n *\n * Defaults: 224×224 RGB input, `float32` normalized with ImageNet mean/std,\n * NCHW layout, batch size 1, softmax applied to the raw output.\n *\n * @example\n * ```typescript\n * const clf = await Classifier.create(\"/models/resnet50.onnx\", {\n *   labels: [\"tench\", \"goldfish\", ...]  // 1000 ImageNet labels\n * });\n * const r = (await clf.predict(\"/images/dog.jpg\"))[0];\n * console.log(r.cls, r.conf, r.name);\n * console.log(r.probs.top5, r.probs.top5conf);\n * ```\n */\nexport class Classifier extends VisionTask {\n    private constructor(\n        session: OrtSession,\n        private readonly _labels: readonly string[],\n        private readonly _names: Readonly<Record<number, string>>,\n        private readonly _inputSize: readonly [number, number],\n        private readonly _mean: readonly [number, number, number],\n        private readonly _std: readonly [number, number, number],\n        private readonly _applySoftmax: boolean,\n    ) {\n        super(session);\n    }\n\n    /** Load the model and resolve labels. */\n    static async create(model: ModelSource, options: ClassifierOptions): Promise<Classifier> {\n        const session = await OrtSession.create(model, options);\n        const labels = resolveLabels(options.labels, { numClasses: options.numClasses });\n        const names: Record<number, string> = {};\n        for (let i = 0; i < labels.length; i++) {\n            names[i] = labels[i] as string;\n        }\n        return new Classifier(\n            session,\n            labels,\n            names,\n            resolveInputSize({\n                graphShape: session.inputShape,\n                requested: options.inputSize,\n                fallback: [224, 224],\n            }),\n            options.mean ?? IMAGENET_MEAN,\n            options.std ?? IMAGENET_STD,\n            options.applySoftmax ?? true,\n        );\n    }\n\n    /** Class labels indexed by class id. */\n    get labels(): readonly string[] {\n        return this._labels;\n    }\n\n    /** Class id → class name dict (matches Ultralytics' `model.names`). */\n    get names(): Readonly<Record<number, string>> {\n        return this._names;\n    }\n\n    /**\n     * The `[width, height]` this task preprocesses to.\n     *\n     * Resolved at creation time from the model's graph when it declares a static\n     * input, so reading it back tells you the resolution inference really runs at\n     * — not merely what was requested.\n     */\n    get inputSize(): readonly [number, number] {\n        return this._inputSize;\n    }\n\n    /** Number of classes the model can predict. */\n    get numClasses(): number {\n        return this._labels.length;\n    }\n\n    /** Alias for {@link predict} (parity with PyTorch `nn.Module.__call__`). */\n    async call(\n        image: ImageInput,\n        options: ClassifierPredictOptions = {},\n    ): Promise<ClassificationResults[]> {\n        return this.predict(image, options);\n    }\n\n    /** Run classification on a single image. */\n    async predict(\n        image: ImageInput,\n        options: ClassifierPredictOptions = {},\n    ): Promise<ClassificationResults[]> {\n        const timer = new SpeedTimer();\n        const path = typeof image === \"string\" ? image : null;\n        const original = await loadImage(image);\n        timer.stage(\"load\");\n        const tensor = this._preprocess(original);\n        timer.stage(\"preprocess\");\n        const outputs = await this._session.run({ [this._session.inputName]: tensor });\n        timer.stage(\"inference\");\n        const firstOutputName = this._session.outputNames[0];\n        if (firstOutputName === undefined) {\n            throw new Error(\"Classifier model has no outputs.\");\n        }\n        const raw = outputs[firstOutputName];\n        if (raw === undefined) {\n            throw new Error(\n                `Classifier model output ${firstOutputName} missing from run() result.`,\n            );\n        }\n        const fullProbs = this._postprocess(raw.data as Float32Array);\n\n        const { indices, values } = topK(fullProbs, options.topK ?? null);\n        const probabilities: ClassProbability[] = [];\n        for (let i = 0; i < indices.length; i++) {\n            const id = indices[i] as number;\n            const className = this._labels[id] ?? `class_${id}`;\n            probabilities.push({\n                classId: id,\n                className,\n                probability: values[i] as number,\n                cls: id,\n                name: className,\n                conf: values[i] as number,\n            });\n        }\n        if (probabilities.length === 0) {\n            throw new Error(\"Classifier produced no probabilities (empty output).\");\n        }\n\n        const top = probabilities[0] as ClassProbability;\n        const result: ClassificationResult = {\n            classId: top.classId,\n            className: top.className,\n            confidence: top.probability,\n            cls: top.classId,\n            name: top.className,\n            conf: top.probability,\n            image: original,\n            probabilities,\n        };\n\n        const orig: readonly [number, number] = [original.height, original.width];\n        const probs = new Probs(fullProbs);\n        timer.stage(\"postprocess\");\n        return [\n            new ClassificationResults(\n                probs,\n                result,\n                this._names,\n                original,\n                orig,\n                path,\n                timer.speed(),\n            ),\n        ];\n    }\n\n    private _preprocess(image: RGBImage): ort.Tensor {\n        const [tw, th] = this._inputSize;\n        const resized = resize(image, tw, th);\n        const normalized = normalize(resized, this._mean, this._std);\n        const chw = toCHW(normalized, resized.width, resized.height, 3);\n        return toFloat32Tensor(chw, [1, 3, resized.height, resized.width]);\n    }\n\n    private _postprocess(raw: Float32Array): Float32Array {\n        return this._applySoftmax ? softmax(raw) : new Float32Array(raw);\n    }\n}\n"],"mappings":"sSAiBA,IAAM,EAAmD,CAAC,KAAO,KAAO,IAAK,EACvE,EAAkD,CAAC,KAAO,KAAO,IAAK,EA4D/D,EAAb,MAAa,UAAmB,EAAA,UAAW,CAGlB,QACA,OACA,WACA,MACA,KACA,cAPrB,YACI,EACA,EACA,EACA,EACA,EACA,EACA,EACF,CACE,MAAM,CAAO,EAPI,KAAA,QAAA,EACA,KAAA,OAAA,EACA,KAAA,WAAA,EACA,KAAA,MAAA,EACA,KAAA,KAAA,EACA,KAAA,cAAA,CAGrB,CAGA,aAAa,OAAO,EAAoB,EAAiD,CACrF,IAAM,EAAU,MAAM,EAAA,WAAW,OAAO,EAAO,CAAO,EAChD,EAAS,EAAA,cAAc,EAAQ,OAAQ,CAAE,WAAY,EAAQ,UAAW,CAAC,EACzE,EAAgC,CAAC,EACvC,IAAK,IAAI,EAAI,EAAG,EAAI,EAAO,OAAQ,IAC/B,EAAM,GAAK,EAAO,GAEtB,OAAO,IAAI,EACP,EACA,EACA,EACA,EAAA,iBAAiB,CACb,WAAY,EAAQ,WACpB,UAAW,EAAQ,UACnB,SAAU,CAAC,IAAK,GAAG,CACvB,CAAC,EACD,EAAQ,MAAQ,EAChB,EAAQ,KAAO,EACf,EAAQ,cAAgB,EAC5B,CACJ,CAGA,IAAI,QAA4B,CAC5B,OAAO,KAAK,OAChB,CAGA,IAAI,OAA0C,CAC1C,OAAO,KAAK,MAChB,CASA,IAAI,WAAuC,CACvC,OAAO,KAAK,UAChB,CAGA,IAAI,YAAqB,CACrB,OAAO,KAAK,QAAQ,MACxB,CAGA,MAAM,KACF,EACA,EAAoC,CAAC,EACL,CAChC,OAAO,KAAK,QAAQ,EAAO,CAAO,CACtC,CAGA,MAAM,QACF,EACA,EAAoC,CAAC,EACL,CAChC,IAAM,EAAQ,IAAI,EAAA,WACZ,EAAO,OAAO,GAAU,SAAW,EAAQ,KAC3C,EAAW,MAAM,EAAA,UAAU,CAAK,EACtC,EAAM,MAAM,MAAM,EAClB,IAAM,EAAS,KAAK,YAAY,CAAQ,EACxC,EAAM,MAAM,YAAY,EACxB,IAAM,EAAU,MAAM,KAAK,SAAS,IAAI,EAAG,KAAK,SAAS,WAAY,CAAO,CAAC,EAC7E,EAAM,MAAM,WAAW,EACvB,IAAM,EAAkB,KAAK,SAAS,YAAY,GAClD,GAAI,IAAoB,IAAA,GACpB,MAAU,MAAM,kCAAkC,EAEtD,IAAM,EAAM,EAAQ,GACpB,GAAI,IAAQ,IAAA,GACR,MAAU,MACN,2BAA2B,EAAgB,4BAC/C,EAEJ,IAAM,EAAY,KAAK,aAAa,EAAI,IAAoB,EAEtD,CAAE,UAAS,UAAW,EAAA,KAAK,EAAW,EAAQ,MAAQ,IAAI,EAC1D,EAAoC,CAAC,EAC3C,IAAK,IAAI,EAAI,EAAG,EAAI,EAAQ,OAAQ,IAAK,CACrC,IAAM,EAAK,EAAQ,GACb,EAAY,KAAK,QAAQ,IAAO,SAAS,IAC/C,EAAc,KAAK,CACf,QAAS,EACT,YACA,YAAa,EAAO,GACpB,IAAK,EACL,KAAM,EACN,KAAM,EAAO,EACjB,CAAC,CACL,CACA,GAAI,EAAc,SAAW,EACzB,MAAU,MAAM,sDAAsD,EAG1E,IAAM,EAAM,EAAc,GACpB,EAA+B,CACjC,QAAS,EAAI,QACb,UAAW,EAAI,UACf,WAAY,EAAI,YAChB,IAAK,EAAI,QACT,KAAM,EAAI,UACV,KAAM,EAAI,YACV,MAAO,EACP,eACJ,EAEM,EAAkC,CAAC,EAAS,OAAQ,EAAS,KAAK,EAClE,EAAQ,IAAI,EAAA,MAAM,CAAS,EAEjC,OADA,EAAM,MAAM,aAAa,EAClB,CACH,IAAI,EAAA,sBACA,EACA,EACA,KAAK,OACL,EACA,EACA,EACA,EAAM,MAAM,CAChB,CACJ,CACJ,CAEA,YAAoB,EAA6B,CAC7C,GAAM,CAAC,EAAI,GAAM,KAAK,WAChB,EAAU,EAAA,OAAO,EAAO,EAAI,CAAE,EAGpC,OAAO,EAAA,gBADK,EAAA,MADO,EAAA,UAAU,EAAS,KAAK,MAAO,KAAK,IACrC,EAAY,EAAQ,MAAO,EAAQ,OAAQ,CACtC,EAAK,CAAC,EAAG,EAAG,EAAQ,OAAQ,EAAQ,KAAK,CAAC,CACrE,CAEA,aAAqB,EAAiC,CAClD,OAAO,KAAK,cAAgB,EAAA,QAAQ,CAAG,EAAI,IAAI,aAAa,CAAG,CACnE,CACJ"}

package/dist/vision/tasks/classifier.js

@@ -1,21 +1,22 @@
 import { SpeedTimer as e } from "../core/timing.js";
 import { ClassificationResults as t, Probs as n } from "../results.js";
 import { resolveLabels as r } from "../labels.js";
-import { OrtSession as i } from "../core/session.js";
-import { loadImage as a } from "../io/image.js";
-import { normalize as o, resize as s, toCHW as c, toFloat32Tensor as l } from "../preprocess/image.js";
-import { softmax as u, topK as d } from "../postprocess/classification.js";
-import { VisionTask as f } from "./base.js";
+import { resolveInputSize as i } from "../core/graph.js";
+import { OrtSession as a } from "../core/session.js";
+import { loadImage as o } from "../io/image.js";
+import { normalize as s, resize as c, toCHW as l, toFloat32Tensor as u } from "../preprocess/image.js";
+import { softmax as d, topK as f } from "../postprocess/classification.js";
+import { VisionTask as p } from "./base.js";
 //#region src/vision/tasks/classifier.ts
-var p = [
+var m = [
 	.485,
 	.456,
 	.406
-], m = [
+], h = [
 	.229,
 	.224,
 	.225
-], h = class h extends f {
+], g = class g extends p {
 	_labels;
 	_names;
 	_inputSize;
@@ -26,9 +27,13 @@ var p = [
 		super(e), this._labels = t, this._names = n, this._inputSize = r, this._mean = i, this._std = a, this._applySoftmax = o;
 	}
 	static async create(e, t) {
-		let n = await i.create(e, t), a = r(t.labels, { numClasses: t.numClasses }), o = {};
-		for (let e = 0; e < a.length; e++) o[e] = a[e];
-		return new h(n, a, o, t.inputSize ?? [224, 224], t.mean ?? p, t.std ?? m, t.applySoftmax ?? !0);
+		let n = await a.create(e, t), o = r(t.labels, { numClasses: t.numClasses }), s = {};
+		for (let e = 0; e < o.length; e++) s[e] = o[e];
+		return new g(n, o, s, i({
+			graphShape: n.inputShape,
+			requested: t.inputSize,
+			fallback: [224, 224]
+		}), t.mean ?? m, t.std ?? h, t.applySoftmax ?? !0);
 	}
 	get labels() {
 		return this._labels;
@@ -36,6 +41,9 @@ var p = [
 	get names() {
 		return this._names;
 	}
+	get inputSize() {
+		return this._inputSize;
+	}
 	get numClasses() {
 		return this._labels.length;
 	}
@@ -43,17 +51,17 @@ var p = [
 		return this.predict(e, t);
 	}
 	async predict(r, i = {}) {
-		let o = new e(), s = typeof r == "string" ? r : null, c = await a(r);
-		o.stage("load");
+		let a = new e(), s = typeof r == "string" ? r : null, c = await o(r);
+		a.stage("load");
 		let l = this._preprocess(c);
-		o.stage("preprocess");
+		a.stage("preprocess");
 		let u = await this._session.run({ [this._session.inputName]: l });
-		o.stage("inference");
-		let f = this._session.outputNames[0];
-		if (f === void 0) throw Error("Classifier model has no outputs.");
-		let p = u[f];
-		if (p === void 0) throw Error(`Classifier model output ${f} missing from run() result.`);
-		let m = this._postprocess(p.data), { indices: h, values: g } = d(m, i.topK ?? null), _ = [];
+		a.stage("inference");
+		let d = this._session.outputNames[0];
+		if (d === void 0) throw Error("Classifier model has no outputs.");
+		let p = u[d];
+		if (p === void 0) throw Error(`Classifier model output ${d} missing from run() result.`);
+		let m = this._postprocess(p.data), { indices: h, values: g } = f(m, i.topK ?? null), _ = [];
 		for (let e = 0; e < h.length; e++) {
 			let t = h[e], n = this._labels[t] ?? `class_${t}`;
 			_.push({
@@ -76,11 +84,11 @@ var p = [
 			image: c,
 			probabilities: _
 		}, b = [c.height, c.width], x = new n(m);
-		return o.stage("postprocess"), [new t(x, y, this._names, c, b, s, o.speed())];
+		return a.stage("postprocess"), [new t(x, y, this._names, c, b, s, a.speed())];
 	}
 	_preprocess(e) {
-		let [t, n] = this._inputSize, r = s(e, t, n);
-		return l(c(o(r, this._mean, this._std), r.width, r.height, 3), [
+		let [t, n] = this._inputSize, r = c(e, t, n);
+		return u(l(s(r, this._mean, this._std), r.width, r.height, 3), [
 			1,
 			3,
 			r.height,
@@ -88,10 +96,10 @@ var p = [
 		]);
 	}
 	_postprocess(e) {
-		return this._applySoftmax ? u(e) : new Float32Array(e);
+		return this._applySoftmax ? d(e) : new Float32Array(e);
 	}
 };
 //#endregion
-export { h as Classifier };
+export { g as Classifier };
 
 //# sourceMappingURL=classifier.js.map

package/dist/vision/tasks/classifier.js.map

@@ -1 +1 @@
-{"version":3,"file":"classifier.js","names":[],"sources":["../../../src/vision/tasks/classifier.ts"],"sourcesContent":["/**\n * Image classification task using ONNX Runtime Web.\n */\n\nimport type * as ort from \"onnxruntime-web\";\n\nimport { type ModelSource, type OrtSessionOptions, OrtSession } from \"../core/session\";\nimport { SpeedTimer } from \"../core/timing\";\nimport { type ImageInput, loadImage } from \"../io/image\";\nimport { type LabelSpec, resolveLabels } from \"../labels\";\nimport { softmax, topK } from \"../postprocess/classification\";\nimport { normalize, resize, toCHW, toFloat32Tensor } from \"../preprocess/image\";\nimport { ClassificationResults, Probs } from \"../results\";\nimport { VisionTask } from \"./base\";\nimport { type ClassProbability, type ClassificationResult, type RGBImage } from \"../types\";\n\nconst IMAGENET_MEAN: readonly [number, number, number] = [0.485, 0.456, 0.406];\nconst IMAGENET_STD: readonly [number, number, number] = [0.229, 0.224, 0.225];\n\nexport interface ClassifierOptions extends OrtSessionOptions {\n    /** Class label spec — see {@link resolveLabels}. */\n    readonly labels: LabelSpec;\n    /**\n     * Number of classes the model can predict. Required when `labels` is `null`\n     * or when you want to validate that the supplied labels match the model.\n     */\n    readonly numClasses?: number;\n    /** Model input `[width, height]` in pixels. Defaults to `[224, 224]`. */\n    readonly inputSize?: readonly [number, number];\n    /** Per-channel RGB mean used for normalization. Defaults to ImageNet. */\n    readonly mean?: readonly [number, number, number];\n    /** Per-channel RGB standard deviation. Defaults to ImageNet. */\n    readonly std?: readonly [number, number, number];\n    /**\n     * If `true` (default), apply softmax to the raw model output. Set to\n     * `false` for models whose final layer already produces a probability\n     * distribution.\n     */\n    readonly applySoftmax?: boolean;\n}\n\nexport interface ClassifierPredictOptions {\n    /**\n     * If set, the per-class probability list in `results[0].result.probabilities`\n     * is truncated to the top-K entries. The bulk `probs` view always exposes\n     * the full vector.\n     */\n    readonly topK?: number;\n}\n\n/**\n * Image classifier wrapping an ONNX model with ImageNet-style preprocessing.\n *\n * `predict()` returns `Promise<ClassificationResults[]>` (length 1 for a\n * single image), mirroring Ultralytics' API. The envelope exposes a `probs`\n * collection (`top1`, `top1conf`, `top5`, `top5conf`, `data`) plus the\n * legacy per-class probability list with names already resolved.\n *\n * Defaults: 224×224 RGB input, `float32` normalized with ImageNet mean/std,\n * NCHW layout, batch size 1, softmax applied to the raw output.\n *\n * @example\n * ```typescript\n * const clf = await Classifier.create(\"/models/resnet50.onnx\", {\n *   labels: [\"tench\", \"goldfish\", ...]  // 1000 ImageNet labels\n * });\n * const r = (await clf.predict(\"/images/dog.jpg\"))[0];\n * console.log(r.cls, r.conf, r.name);\n * console.log(r.probs.top5, r.probs.top5conf);\n * ```\n */\nexport class Classifier extends VisionTask {\n    private constructor(\n        session: OrtSession,\n        private readonly _labels: readonly string[],\n        private readonly _names: Readonly<Record<number, string>>,\n        private readonly _inputSize: readonly [number, number],\n        private readonly _mean: readonly [number, number, number],\n        private readonly _std: readonly [number, number, number],\n        private readonly _applySoftmax: boolean,\n    ) {\n        super(session);\n    }\n\n    /** Load the model and resolve labels. */\n    static async create(model: ModelSource, options: ClassifierOptions): Promise<Classifier> {\n        const session = await OrtSession.create(model, options);\n        const labels = resolveLabels(options.labels, { numClasses: options.numClasses });\n        const names: Record<number, string> = {};\n        for (let i = 0; i < labels.length; i++) {\n            names[i] = labels[i] as string;\n        }\n        return new Classifier(\n            session,\n            labels,\n            names,\n            options.inputSize ?? [224, 224],\n            options.mean ?? IMAGENET_MEAN,\n            options.std ?? IMAGENET_STD,\n            options.applySoftmax ?? true,\n        );\n    }\n\n    /** Class labels indexed by class id. */\n    get labels(): readonly string[] {\n        return this._labels;\n    }\n\n    /** Class id → class name dict (matches Ultralytics' `model.names`). */\n    get names(): Readonly<Record<number, string>> {\n        return this._names;\n    }\n\n    /** Number of classes the model can predict. */\n    get numClasses(): number {\n        return this._labels.length;\n    }\n\n    /** Alias for {@link predict} (parity with PyTorch `nn.Module.__call__`). */\n    async call(\n        image: ImageInput,\n        options: ClassifierPredictOptions = {},\n    ): Promise<ClassificationResults[]> {\n        return this.predict(image, options);\n    }\n\n    /** Run classification on a single image. */\n    async predict(\n        image: ImageInput,\n        options: ClassifierPredictOptions = {},\n    ): Promise<ClassificationResults[]> {\n        const timer = new SpeedTimer();\n        const path = typeof image === \"string\" ? image : null;\n        const original = await loadImage(image);\n        timer.stage(\"load\");\n        const tensor = this._preprocess(original);\n        timer.stage(\"preprocess\");\n        const outputs = await this._session.run({ [this._session.inputName]: tensor });\n        timer.stage(\"inference\");\n        const firstOutputName = this._session.outputNames[0];\n        if (firstOutputName === undefined) {\n            throw new Error(\"Classifier model has no outputs.\");\n        }\n        const raw = outputs[firstOutputName];\n        if (raw === undefined) {\n            throw new Error(\n                `Classifier model output ${firstOutputName} missing from run() result.`,\n            );\n        }\n        const fullProbs = this._postprocess(raw.data as Float32Array);\n\n        const { indices, values } = topK(fullProbs, options.topK ?? null);\n        const probabilities: ClassProbability[] = [];\n        for (let i = 0; i < indices.length; i++) {\n            const id = indices[i] as number;\n            const className = this._labels[id] ?? `class_${id}`;\n            probabilities.push({\n                classId: id,\n                className,\n                probability: values[i] as number,\n                cls: id,\n                name: className,\n                conf: values[i] as number,\n            });\n        }\n        if (probabilities.length === 0) {\n            throw new Error(\"Classifier produced no probabilities (empty output).\");\n        }\n\n        const top = probabilities[0] as ClassProbability;\n        const result: ClassificationResult = {\n            classId: top.classId,\n            className: top.className,\n            confidence: top.probability,\n            cls: top.classId,\n            name: top.className,\n            conf: top.probability,\n            image: original,\n            probabilities,\n        };\n\n        const orig: readonly [number, number] = [original.height, original.width];\n        const probs = new Probs(fullProbs);\n        timer.stage(\"postprocess\");\n        return [\n            new ClassificationResults(\n                probs,\n                result,\n                this._names,\n                original,\n                orig,\n                path,\n                timer.speed(),\n            ),\n        ];\n    }\n\n    private _preprocess(image: RGBImage): ort.Tensor {\n        const [tw, th] = this._inputSize;\n        const resized = resize(image, tw, th);\n        const normalized = normalize(resized, this._mean, this._std);\n        const chw = toCHW(normalized, resized.width, resized.height, 3);\n        return toFloat32Tensor(chw, [1, 3, resized.height, resized.width]);\n    }\n\n    private _postprocess(raw: Float32Array): Float32Array {\n        return this._applySoftmax ? softmax(raw) : new Float32Array(raw);\n    }\n}\n"],"mappings":";;;;;;;;;AAgBA,IAAM,IAAmD;CAAC;CAAO;CAAO;AAAK,GACvE,IAAkD;CAAC;CAAO;CAAO;AAAK,GAsD/D,IAAb,MAAa,UAAmB,EAAW;CAGlB;CACA;CACA;CACA;CACA;CACA;CAPrB,YACI,GACA,GACA,GACA,GACA,GACA,GACA,GACF;EADmB,AAEjB,MAAM,CAAO,GAPI,KAAA,UAAA,GACA,KAAA,SAAA,GACA,KAAA,aAAA,GACA,KAAA,QAAA,GACA,KAAA,OAAA,GACA,KAAA,gBAAA;CAGrB;CAGA,aAAa,OAAO,GAAoB,GAAiD;EACrF,IAAM,IAAU,MAAM,EAAW,OAAO,GAAO,CAAO,GAChD,IAAS,EAAc,EAAQ,QAAQ,EAAE,YAAY,EAAQ,WAAW,CAAC,GACzE,IAAgC,CAAC;EACvC,KAAK,IAAI,IAAI,GAAG,IAAI,EAAO,QAAQ,KAC/B,EAAM,KAAK,EAAO;EAEtB,OAAO,IAAI,EACP,GACA,GACA,GACA,EAAQ,aAAa,CAAC,KAAK,GAAG,GAC9B,EAAQ,QAAQ,GAChB,EAAQ,OAAO,GACf,EAAQ,gBAAgB,EAC5B;CACJ;CAGA,IAAI,SAA4B;EAC5B,OAAO,KAAK;CAChB;CAGA,IAAI,QAA0C;EAC1C,OAAO,KAAK;CAChB;CAGA,IAAI,aAAqB;EACrB,OAAO,KAAK,QAAQ;CACxB;CAGA,MAAM,KACF,GACA,IAAoC,CAAC,GACL;EAChC,OAAO,KAAK,QAAQ,GAAO,CAAO;CACtC;CAGA,MAAM,QACF,GACA,IAAoC,CAAC,GACL;EAChC,IAAM,IAAQ,IAAI,EAAW,GACvB,IAAO,OAAO,KAAU,WAAW,IAAQ,MAC3C,IAAW,MAAM,EAAU,CAAK;EACtC,EAAM,MAAM,MAAM;EAClB,IAAM,IAAS,KAAK,YAAY,CAAQ;EACxC,EAAM,MAAM,YAAY;EACxB,IAAM,IAAU,MAAM,KAAK,SAAS,IAAI,GAAG,KAAK,SAAS,YAAY,EAAO,CAAC;EAC7E,EAAM,MAAM,WAAW;EACvB,IAAM,IAAkB,KAAK,SAAS,YAAY;EAClD,IAAI,MAAoB,KAAA,GACpB,MAAU,MAAM,kCAAkC;EAEtD,IAAM,IAAM,EAAQ;EACpB,IAAI,MAAQ,KAAA,GACR,MAAU,MACN,2BAA2B,EAAgB,4BAC/C;EAEJ,IAAM,IAAY,KAAK,aAAa,EAAI,IAAoB,GAEtD,EAAE,YAAS,cAAW,EAAK,GAAW,EAAQ,QAAQ,IAAI,GAC1D,IAAoC,CAAC;EAC3C,KAAK,IAAI,IAAI,GAAG,IAAI,EAAQ,QAAQ,KAAK;GACrC,IAAM,IAAK,EAAQ,IACb,IAAY,KAAK,QAAQ,MAAO,SAAS;GAC/C,EAAc,KAAK;IACf,SAAS;IACT;IACA,aAAa,EAAO;IACpB,KAAK;IACL,MAAM;IACN,MAAM,EAAO;GACjB,CAAC;EACL;EACA,IAAI,EAAc,WAAW,GACzB,MAAU,MAAM,sDAAsD;EAG1E,IAAM,IAAM,EAAc,IACpB,IAA+B;GACjC,SAAS,EAAI;GACb,WAAW,EAAI;GACf,YAAY,EAAI;GAChB,KAAK,EAAI;GACT,MAAM,EAAI;GACV,MAAM,EAAI;GACV,OAAO;GACP;EACJ,GAEM,IAAkC,CAAC,EAAS,QAAQ,EAAS,KAAK,GAClE,IAAQ,IAAI,EAAM,CAAS;EAEjC,OADA,EAAM,MAAM,aAAa,GAClB,CACH,IAAI,EACA,GACA,GACA,KAAK,QACL,GACA,GACA,GACA,EAAM,MAAM,CAChB,CACJ;CACJ;CAEA,YAAoB,GAA6B;EAC7C,IAAM,CAAC,GAAI,KAAM,KAAK,YAChB,IAAU,EAAO,GAAO,GAAI,CAAE;EAGpC,OAAO,EADK,EADO,EAAU,GAAS,KAAK,OAAO,KAAK,IACrC,GAAY,EAAQ,OAAO,EAAQ,QAAQ,CACtC,GAAK;GAAC;GAAG;GAAG,EAAQ;GAAQ,EAAQ;EAAK,CAAC;CACrE;CAEA,aAAqB,GAAiC;EAClD,OAAO,KAAK,gBAAgB,EAAQ,CAAG,IAAI,IAAI,aAAa,CAAG;CACnE;AACJ"}
+{"version":3,"file":"classifier.js","names":[],"sources":["../../../src/vision/tasks/classifier.ts"],"sourcesContent":["/**\n * Image classification task using ONNX Runtime Web.\n */\n\nimport type * as ort from \"onnxruntime-web\";\n\nimport { type ModelSource, type OrtSessionOptions, OrtSession } from \"../core/session\";\nimport { SpeedTimer } from \"../core/timing\";\nimport { resolveInputSize } from \"../core/graph\";\nimport { type ImageInput, loadImage } from \"../io/image\";\nimport { type LabelSpec, resolveLabels } from \"../labels\";\nimport { softmax, topK } from \"../postprocess/classification\";\nimport { normalize, resize, toCHW, toFloat32Tensor } from \"../preprocess/image\";\nimport { ClassificationResults, Probs } from \"../results\";\nimport { VisionTask } from \"./base\";\nimport { type ClassProbability, type ClassificationResult, type RGBImage } from \"../types\";\n\nconst IMAGENET_MEAN: readonly [number, number, number] = [0.485, 0.456, 0.406];\nconst IMAGENET_STD: readonly [number, number, number] = [0.229, 0.224, 0.225];\n\nexport interface ClassifierOptions extends OrtSessionOptions {\n    /** Class label spec — see {@link resolveLabels}. */\n    readonly labels: LabelSpec;\n    /**\n     * Number of classes the model can predict. Required when `labels` is `null`\n     * or when you want to validate that the supplied labels match the model.\n     */\n    readonly numClasses?: number;\n    /**\n     * Model input `[width, height]` in pixels.\n     *\n     * Only used when the model's graph leaves its spatial axes dynamic: a graph\n     * that declares a static size always wins, since that is the only shape ONNX\n     * Runtime will accept. Defaults to `[224, 224]`.\n     */\n    readonly inputSize?: readonly [number, number];\n    /** Per-channel RGB mean used for normalization. Defaults to ImageNet. */\n    readonly mean?: readonly [number, number, number];\n    /** Per-channel RGB standard deviation. Defaults to ImageNet. */\n    readonly std?: readonly [number, number, number];\n    /**\n     * If `true` (default), apply softmax to the raw model output. Set to\n     * `false` for models whose final layer already produces a probability\n     * distribution.\n     */\n    readonly applySoftmax?: boolean;\n}\n\nexport interface ClassifierPredictOptions {\n    /**\n     * If set, the per-class probability list in `results[0].result.probabilities`\n     * is truncated to the top-K entries. The bulk `probs` view always exposes\n     * the full vector.\n     */\n    readonly topK?: number;\n}\n\n/**\n * Image classifier wrapping an ONNX model with ImageNet-style preprocessing.\n *\n * `predict()` returns `Promise<ClassificationResults[]>` (length 1 for a\n * single image), mirroring Ultralytics' API. The envelope exposes a `probs`\n * collection (`top1`, `top1conf`, `top5`, `top5conf`, `data`) plus the\n * legacy per-class probability list with names already resolved.\n *\n * Defaults: 224×224 RGB input, `float32` normalized with ImageNet mean/std,\n * NCHW layout, batch size 1, softmax applied to the raw output.\n *\n * @example\n * ```typescript\n * const clf = await Classifier.create(\"/models/resnet50.onnx\", {\n *   labels: [\"tench\", \"goldfish\", ...]  // 1000 ImageNet labels\n * });\n * const r = (await clf.predict(\"/images/dog.jpg\"))[0];\n * console.log(r.cls, r.conf, r.name);\n * console.log(r.probs.top5, r.probs.top5conf);\n * ```\n */\nexport class Classifier extends VisionTask {\n    private constructor(\n        session: OrtSession,\n        private readonly _labels: readonly string[],\n        private readonly _names: Readonly<Record<number, string>>,\n        private readonly _inputSize: readonly [number, number],\n        private readonly _mean: readonly [number, number, number],\n        private readonly _std: readonly [number, number, number],\n        private readonly _applySoftmax: boolean,\n    ) {\n        super(session);\n    }\n\n    /** Load the model and resolve labels. */\n    static async create(model: ModelSource, options: ClassifierOptions): Promise<Classifier> {\n        const session = await OrtSession.create(model, options);\n        const labels = resolveLabels(options.labels, { numClasses: options.numClasses });\n        const names: Record<number, string> = {};\n        for (let i = 0; i < labels.length; i++) {\n            names[i] = labels[i] as string;\n        }\n        return new Classifier(\n            session,\n            labels,\n            names,\n            resolveInputSize({\n                graphShape: session.inputShape,\n                requested: options.inputSize,\n                fallback: [224, 224],\n            }),\n            options.mean ?? IMAGENET_MEAN,\n            options.std ?? IMAGENET_STD,\n            options.applySoftmax ?? true,\n        );\n    }\n\n    /** Class labels indexed by class id. */\n    get labels(): readonly string[] {\n        return this._labels;\n    }\n\n    /** Class id → class name dict (matches Ultralytics' `model.names`). */\n    get names(): Readonly<Record<number, string>> {\n        return this._names;\n    }\n\n    /**\n     * The `[width, height]` this task preprocesses to.\n     *\n     * Resolved at creation time from the model's graph when it declares a static\n     * input, so reading it back tells you the resolution inference really runs at\n     * — not merely what was requested.\n     */\n    get inputSize(): readonly [number, number] {\n        return this._inputSize;\n    }\n\n    /** Number of classes the model can predict. */\n    get numClasses(): number {\n        return this._labels.length;\n    }\n\n    /** Alias for {@link predict} (parity with PyTorch `nn.Module.__call__`). */\n    async call(\n        image: ImageInput,\n        options: ClassifierPredictOptions = {},\n    ): Promise<ClassificationResults[]> {\n        return this.predict(image, options);\n    }\n\n    /** Run classification on a single image. */\n    async predict(\n        image: ImageInput,\n        options: ClassifierPredictOptions = {},\n    ): Promise<ClassificationResults[]> {\n        const timer = new SpeedTimer();\n        const path = typeof image === \"string\" ? image : null;\n        const original = await loadImage(image);\n        timer.stage(\"load\");\n        const tensor = this._preprocess(original);\n        timer.stage(\"preprocess\");\n        const outputs = await this._session.run({ [this._session.inputName]: tensor });\n        timer.stage(\"inference\");\n        const firstOutputName = this._session.outputNames[0];\n        if (firstOutputName === undefined) {\n            throw new Error(\"Classifier model has no outputs.\");\n        }\n        const raw = outputs[firstOutputName];\n        if (raw === undefined) {\n            throw new Error(\n                `Classifier model output ${firstOutputName} missing from run() result.`,\n            );\n        }\n        const fullProbs = this._postprocess(raw.data as Float32Array);\n\n        const { indices, values } = topK(fullProbs, options.topK ?? null);\n        const probabilities: ClassProbability[] = [];\n        for (let i = 0; i < indices.length; i++) {\n            const id = indices[i] as number;\n            const className = this._labels[id] ?? `class_${id}`;\n            probabilities.push({\n                classId: id,\n                className,\n                probability: values[i] as number,\n                cls: id,\n                name: className,\n                conf: values[i] as number,\n            });\n        }\n        if (probabilities.length === 0) {\n            throw new Error(\"Classifier produced no probabilities (empty output).\");\n        }\n\n        const top = probabilities[0] as ClassProbability;\n        const result: ClassificationResult = {\n            classId: top.classId,\n            className: top.className,\n            confidence: top.probability,\n            cls: top.classId,\n            name: top.className,\n            conf: top.probability,\n            image: original,\n            probabilities,\n        };\n\n        const orig: readonly [number, number] = [original.height, original.width];\n        const probs = new Probs(fullProbs);\n        timer.stage(\"postprocess\");\n        return [\n            new ClassificationResults(\n                probs,\n                result,\n                this._names,\n                original,\n                orig,\n                path,\n                timer.speed(),\n            ),\n        ];\n    }\n\n    private _preprocess(image: RGBImage): ort.Tensor {\n        const [tw, th] = this._inputSize;\n        const resized = resize(image, tw, th);\n        const normalized = normalize(resized, this._mean, this._std);\n        const chw = toCHW(normalized, resized.width, resized.height, 3);\n        return toFloat32Tensor(chw, [1, 3, resized.height, resized.width]);\n    }\n\n    private _postprocess(raw: Float32Array): Float32Array {\n        return this._applySoftmax ? softmax(raw) : new Float32Array(raw);\n    }\n}\n"],"mappings":";;;;;;;;;;AAiBA,IAAM,IAAmD;CAAC;CAAO;CAAO;AAAK,GACvE,IAAkD;CAAC;CAAO;CAAO;AAAK,GA4D/D,IAAb,MAAa,UAAmB,EAAW;CAGlB;CACA;CACA;CACA;CACA;CACA;CAPrB,YACI,GACA,GACA,GACA,GACA,GACA,GACA,GACF;EADmB,AAEjB,MAAM,CAAO,GAPI,KAAA,UAAA,GACA,KAAA,SAAA,GACA,KAAA,aAAA,GACA,KAAA,QAAA,GACA,KAAA,OAAA,GACA,KAAA,gBAAA;CAGrB;CAGA,aAAa,OAAO,GAAoB,GAAiD;EACrF,IAAM,IAAU,MAAM,EAAW,OAAO,GAAO,CAAO,GAChD,IAAS,EAAc,EAAQ,QAAQ,EAAE,YAAY,EAAQ,WAAW,CAAC,GACzE,IAAgC,CAAC;EACvC,KAAK,IAAI,IAAI,GAAG,IAAI,EAAO,QAAQ,KAC/B,EAAM,KAAK,EAAO;EAEtB,OAAO,IAAI,EACP,GACA,GACA,GACA,EAAiB;GACb,YAAY,EAAQ;GACpB,WAAW,EAAQ;GACnB,UAAU,CAAC,KAAK,GAAG;EACvB,CAAC,GACD,EAAQ,QAAQ,GAChB,EAAQ,OAAO,GACf,EAAQ,gBAAgB,EAC5B;CACJ;CAGA,IAAI,SAA4B;EAC5B,OAAO,KAAK;CAChB;CAGA,IAAI,QAA0C;EAC1C,OAAO,KAAK;CAChB;CASA,IAAI,YAAuC;EACvC,OAAO,KAAK;CAChB;CAGA,IAAI,aAAqB;EACrB,OAAO,KAAK,QAAQ;CACxB;CAGA,MAAM,KACF,GACA,IAAoC,CAAC,GACL;EAChC,OAAO,KAAK,QAAQ,GAAO,CAAO;CACtC;CAGA,MAAM,QACF,GACA,IAAoC,CAAC,GACL;EAChC,IAAM,IAAQ,IAAI,EAAW,GACvB,IAAO,OAAO,KAAU,WAAW,IAAQ,MAC3C,IAAW,MAAM,EAAU,CAAK;EACtC,EAAM,MAAM,MAAM;EAClB,IAAM,IAAS,KAAK,YAAY,CAAQ;EACxC,EAAM,MAAM,YAAY;EACxB,IAAM,IAAU,MAAM,KAAK,SAAS,IAAI,GAAG,KAAK,SAAS,YAAY,EAAO,CAAC;EAC7E,EAAM,MAAM,WAAW;EACvB,IAAM,IAAkB,KAAK,SAAS,YAAY;EAClD,IAAI,MAAoB,KAAA,GACpB,MAAU,MAAM,kCAAkC;EAEtD,IAAM,IAAM,EAAQ;EACpB,IAAI,MAAQ,KAAA,GACR,MAAU,MACN,2BAA2B,EAAgB,4BAC/C;EAEJ,IAAM,IAAY,KAAK,aAAa,EAAI,IAAoB,GAEtD,EAAE,YAAS,cAAW,EAAK,GAAW,EAAQ,QAAQ,IAAI,GAC1D,IAAoC,CAAC;EAC3C,KAAK,IAAI,IAAI,GAAG,IAAI,EAAQ,QAAQ,KAAK;GACrC,IAAM,IAAK,EAAQ,IACb,IAAY,KAAK,QAAQ,MAAO,SAAS;GAC/C,EAAc,KAAK;IACf,SAAS;IACT;IACA,aAAa,EAAO;IACpB,KAAK;IACL,MAAM;IACN,MAAM,EAAO;GACjB,CAAC;EACL;EACA,IAAI,EAAc,WAAW,GACzB,MAAU,MAAM,sDAAsD;EAG1E,IAAM,IAAM,EAAc,IACpB,IAA+B;GACjC,SAAS,EAAI;GACb,WAAW,EAAI;GACf,YAAY,EAAI;GAChB,KAAK,EAAI;GACT,MAAM,EAAI;GACV,MAAM,EAAI;GACV,OAAO;GACP;EACJ,GAEM,IAAkC,CAAC,EAAS,QAAQ,EAAS,KAAK,GAClE,IAAQ,IAAI,EAAM,CAAS;EAEjC,OADA,EAAM,MAAM,aAAa,GAClB,CACH,IAAI,EACA,GACA,GACA,KAAK,QACL,GACA,GACA,GACA,EAAM,MAAM,CAChB,CACJ;CACJ;CAEA,YAAoB,GAA6B;EAC7C,IAAM,CAAC,GAAI,KAAM,KAAK,YAChB,IAAU,EAAO,GAAO,GAAI,CAAE;EAGpC,OAAO,EADK,EADO,EAAU,GAAS,KAAK,OAAO,KAAK,IACrC,GAAY,EAAQ,OAAO,EAAQ,QAAQ,CACtC,GAAK;GAAC;GAAG;GAAG,EAAQ;GAAQ,EAAQ;EAAK,CAAC;CACrE;CAEA,aAAqB,GAAiC;EAClD,OAAO,KAAK,gBAAgB,EAAQ,CAAG,IAAI,IAAI,aAAa,CAAG;CACnE;AACJ"}

package/dist/vision/tasks/detector.cjs

@@ -1,2 +1,2 @@
-const e=require("../types.cjs"),t=require("../core/timing.cjs"),n=require("../results.cjs"),r=require("../labels.cjs"),i=require("../core/session.cjs"),a=require("../io/image.cjs"),o=require("../preprocess/image.cjs"),s=require("../postprocess/detection.cjs"),c=require("./base.cjs");var l=class l extends c.VisionTask{_head;_labels;_names;_inputSize;_confThreshold;_iouThreshold;_maxDetections;constructor(e,t,n,r,i,a,o,s){super(e),this._head=t,this._labels=n,this._names=r,this._inputSize=i,this._confThreshold=a,this._iouThreshold=o,this._maxDetections=s}static async create(e,t={}){let n=t.head??`yolo`;if(n!==`yolo`)throw Error(`Unsupported detector head '${n}'. Supported: 'yolo'.`);let a=await i.OrtSession.create(e,t),o=r.resolveLabels(t.labels??`coco`,{numClasses:t.numClasses}),s={};for(let e=0;e<o.length;e++)s[e]=o[e];return new l(a,n,o,s,t.inputSize??[640,640],t.confThreshold??.25,t.iouThreshold??.45,t.maxDetections??300)}get head(){return this._head}get labels(){return this._labels}get names(){return this._names}get numClasses(){return this._labels.length}async call(e,t={}){return this.predict(e,t)}async predict(e,r={}){let i=new t.SpeedTimer,o=typeof e==`string`?e:null,c=await a.loadImage(e);i.stage(`load`);let{tensor:l,scale:u,padLeft:d,padTop:f}=this._preprocess(c);i.stage(`preprocess`);let p=await this._session.run({[this._session.inputName]:l});i.stage(`inference`);let m=this._session.outputNames[0];if(m===void 0)throw Error(`Detector model has no outputs.`);let h=p[m];if(h===void 0)throw Error(`Detector model output ${m} missing from run() result.`);let g=s.decodeYolo(h.data,h.dims,{originalWidth:c.width,originalHeight:c.height,padLeft:d,padTop:f,scale:u,confThreshold:r.confThreshold??this._confThreshold,iouThreshold:r.iouThreshold??this._iouThreshold,maxDetections:this._maxDetections}),_=(r.classes===void 0?g:(()=>{let e=new Set(r.classes);return g.filter(t=>e.has(t.classId))})()).map(e=>this._buildResult(c,e.bbox,e.classId,e.confidence)),v=[c.height,c.width],y=this._buildBoxes(_,v);return i.stage(`postprocess`),[new n.DetectionResults(y,_,this._names,c,v,o,i.speed())]}_preprocess(e){let[t,n]=this._inputSize,r=o.letterbox(e,t,n);return{tensor:o.toFloat32Tensor(o.toCHW(o.toFloat32(r.image),r.image.width,r.image.height,3),[1,3,r.image.height,r.image.width]),scale:r.scale,padLeft:r.padLeft,padTop:r.padTop}}_buildResult(t,n,r,i){let[a,o,s,c]=n.asIntXyxy(),l=Math.max(0,a),u=Math.max(0,o),d=Math.min(t.width,s),f=Math.min(t.height,c),p;if(d>l&&f>u){let n=d-l,r=f-u,i=new Uint8Array(n*r*3);for(let e=0;e<r;e++){let r=((u+e)*t.width+l)*3;i.set(t.data.subarray(r,r+n*3),e*n*3)}p=new e.RGBImage(i,n,r)}else p=new e.RGBImage(new Uint8Array,0,0);let m=this._names[r]??`class_${r}`;return{classId:r,className:m,confidence:i,bbox:n,cls:r,name:m,conf:i,box:n,croppedImage:p}}_buildBoxes(e,t){let r=e.length,i=new Float32Array(r*4),a=new Int32Array(r),o=new Float32Array(r);for(let t=0;t<r;t++){let n=e[t];i[t*4]=n.bbox.x1,i[t*4+1]=n.bbox.y1,i[t*4+2]=n.bbox.x2,i[t*4+3]=n.bbox.y2,a[t]=n.classId,o[t]=n.confidence}return new n.Boxes(i,a,o,t)}};exports.Detector=l;
+const e=require("../types.cjs"),t=require("../core/timing.cjs"),n=require("../results.cjs"),r=require("../labels.cjs"),i=require("../core/graph.cjs"),a=require("../core/session.cjs"),o=require("../io/image.cjs"),s=require("../preprocess/image.cjs"),c=require("../postprocess/detection.cjs"),l=require("./base.cjs");var u=class u extends l.VisionTask{_head;_labels;_names;_inputSize;_confThreshold;_iouThreshold;_maxDetections;constructor(e,t,n,r,i,a,o,s){super(e),this._head=t,this._labels=n,this._names=r,this._inputSize=i,this._confThreshold=a,this._iouThreshold=o,this._maxDetections=s}static async create(e,t={}){let n=t.head??`yolo`;if(n!==`yolo`)throw Error(`Unsupported detector head '${n}'. Supported: 'yolo'.`);let o=await a.OrtSession.create(e,t),s=r.resolveLabels(t.labels??`coco`,{numClasses:t.numClasses}),c={};for(let e=0;e<s.length;e++)c[e]=s[e];return new u(o,n,s,c,i.resolveInputSize({graphShape:o.inputShape,requested:t.inputSize,fallback:[640,640]}),t.confThreshold??.25,t.iouThreshold??.45,t.maxDetections??300)}get head(){return this._head}get labels(){return this._labels}get names(){return this._names}get inputSize(){return this._inputSize}get numClasses(){return this._labels.length}async call(e,t={}){return this.predict(e,t)}async predict(e,r={}){let i=new t.SpeedTimer,a=typeof e==`string`?e:null,s=await o.loadImage(e);i.stage(`load`);let{tensor:l,scale:u,padLeft:d,padTop:f}=this._preprocess(s);i.stage(`preprocess`);let p=await this._session.run({[this._session.inputName]:l});i.stage(`inference`);let m=this._session.outputNames[0];if(m===void 0)throw Error(`Detector model has no outputs.`);let h=p[m];if(h===void 0)throw Error(`Detector model output ${m} missing from run() result.`);let g=c.decodeYolo(h.data,h.dims,{originalWidth:s.width,originalHeight:s.height,padLeft:d,padTop:f,scale:u,confThreshold:r.confThreshold??this._confThreshold,iouThreshold:r.iouThreshold??this._iouThreshold,maxDetections:this._maxDetections}),_=(r.classes===void 0?g:(()=>{let e=new Set(r.classes);return g.filter(t=>e.has(t.classId))})()).map(e=>this._buildResult(s,e.bbox,e.classId,e.confidence)),v=[s.height,s.width],y=this._buildBoxes(_,v);return i.stage(`postprocess`),[new n.DetectionResults(y,_,this._names,s,v,a,i.speed())]}_preprocess(e){let[t,n]=this._inputSize,r=s.letterbox(e,t,n);return{tensor:s.toFloat32Tensor(s.toCHW(s.toFloat32(r.image),r.image.width,r.image.height,3),[1,3,r.image.height,r.image.width]),scale:r.scale,padLeft:r.padLeft,padTop:r.padTop}}_buildResult(t,n,r,i){let[a,o,s,c]=n.asIntXyxy(),l=Math.max(0,a),u=Math.max(0,o),d=Math.min(t.width,s),f=Math.min(t.height,c),p;if(d>l&&f>u){let n=d-l,r=f-u,i=new Uint8Array(n*r*3);for(let e=0;e<r;e++){let r=((u+e)*t.width+l)*3;i.set(t.data.subarray(r,r+n*3),e*n*3)}p=new e.RGBImage(i,n,r)}else p=new e.RGBImage(new Uint8Array,0,0);let m=this._names[r]??`class_${r}`;return{classId:r,className:m,confidence:i,bbox:n,cls:r,name:m,conf:i,box:n,croppedImage:p}}_buildBoxes(e,t){let r=e.length,i=new Float32Array(r*4),a=new Int32Array(r),o=new Float32Array(r);for(let t=0;t<r;t++){let n=e[t];i[t*4]=n.bbox.x1,i[t*4+1]=n.bbox.y1,i[t*4+2]=n.bbox.x2,i[t*4+3]=n.bbox.y2,a[t]=n.classId,o[t]=n.confidence}return new n.Boxes(i,a,o,t)}};exports.Detector=u;
 //# sourceMappingURL=detector.cjs.map