@camstack/addon-vision 0.1.7 → 0.1.10

package/dist/addons/object-detection/index.js

@@ -5,9 +5,6 @@ var __getOwnPropDesc = Object.getOwnPropertyDescriptor;
 var __getOwnPropNames = Object.getOwnPropertyNames;
 var __getProtoOf = Object.getPrototypeOf;
 var __hasOwnProp = Object.prototype.hasOwnProperty;
-var __commonJS = (cb, mod) => function __require() {
-  return mod || (0, cb[__getOwnPropNames(cb)[0]])((mod = { exports: {} }).exports, mod), mod.exports;
-};
 var __export = (target, all) => {
   for (var name in all)
     __defProp(target, name, { get: all[name], enumerable: true });
@@ -30,1349 +27,1147 @@ var __toESM = (mod, isNodeMode, target) => (target = mod != null ? __create(__ge
 ));
 var __toCommonJS = (mod) => __copyProps(__defProp({}, "__esModule", { value: true }), mod);
 
-// src/catalogs/object-detection-models.js
-var require_object_detection_models = __commonJS({
-  "src/catalogs/object-detection-models.js"(exports2) {
-    "use strict";
-    Object.defineProperty(exports2, "__esModule", { value: true });
-    exports2.OBJECT_DETECTION_MODELS = exports2.MLPACKAGE_FILES = void 0;
-    var types_1 = require("@camstack/types");
-    var HF_REPO = "camstack/camstack-models";
-    exports2.MLPACKAGE_FILES = [
-      "Manifest.json",
-      "Data/com.apple.CoreML/model.mlmodel",
-      "Data/com.apple.CoreML/weights/weight.bin"
-    ];
-    exports2.OBJECT_DETECTION_MODELS = [
-      // ── YOLOv8 ──────────────────────────────────────────────────────
-      {
-        id: "yolov8n",
-        name: "YOLOv8 Nano",
-        description: "YOLOv8 Nano \u2014 fastest, smallest object detection model",
-        inputSize: { width: 640, height: 640 },
-        labels: types_1.COCO_80_LABELS,
-        formats: {
-          onnx: {
-            url: (0, types_1.hfModelUrl)(HF_REPO, "objectDetection/yolov8/onnx/camstack-yolov8n.onnx"),
-            sizeMB: 12
-          },
-          coreml: {
-            url: (0, types_1.hfModelUrl)(HF_REPO, "objectDetection/yolov8/coreml/camstack-yolov8n.mlpackage"),
-            sizeMB: 6,
-            isDirectory: true,
-            files: exports2.MLPACKAGE_FILES,
-            runtimes: ["python"]
-          },
-          openvino: {
-            url: (0, types_1.hfModelUrl)(HF_REPO, "objectDetection/yolov8/openvino/camstack-yolov8n.xml"),
-            sizeMB: 7,
-            runtimes: ["python"]
-          },
-          tflite: {
-            url: (0, types_1.hfModelUrl)(HF_REPO, "objectDetection/yolov8/tflite/camstack-yolov8n_float32.tflite"),
-            sizeMB: 12,
-            runtimes: ["python"]
-          }
-        }
+// src/addons/object-detection/index.ts
+var object_detection_exports = {};
+__export(object_detection_exports, {
+  default: () => ObjectDetectionAddon
+});
+module.exports = __toCommonJS(object_detection_exports);
+
+// src/catalogs/object-detection-models.ts
+var import_types = require("@camstack/types");
+var HF_REPO = "camstack/camstack-models";
+var MLPACKAGE_FILES = [
+  "Manifest.json",
+  "Data/com.apple.CoreML/model.mlmodel",
+  "Data/com.apple.CoreML/weights/weight.bin"
+];
+var OBJECT_DETECTION_MODELS = [
+  // ── YOLOv8 ──────────────────────────────────────────────────────
+  {
+    id: "yolov8n",
+    name: "YOLOv8 Nano",
+    description: "YOLOv8 Nano \u2014 fastest, smallest object detection model",
+    inputSize: { width: 640, height: 640 },
+    labels: import_types.COCO_80_LABELS,
+    formats: {
+      onnx: {
+        url: (0, import_types.hfModelUrl)(HF_REPO, "objectDetection/yolov8/onnx/camstack-yolov8n.onnx"),
+        sizeMB: 12
       },
-      {
-        id: "yolov8s",
-        name: "YOLOv8 Small",
-        description: "YOLOv8 Small \u2014 balanced speed and accuracy",
-        inputSize: { width: 640, height: 640 },
-        labels: types_1.COCO_80_LABELS,
-        formats: {
-          onnx: {
-            url: (0, types_1.hfModelUrl)(HF_REPO, "objectDetection/yolov8/onnx/camstack-yolov8s.onnx"),
-            sizeMB: 43
-          },
-          coreml: {
-            url: (0, types_1.hfModelUrl)(HF_REPO, "objectDetection/yolov8/coreml/camstack-yolov8s.mlpackage"),
-            sizeMB: 21,
-            isDirectory: true,
-            files: exports2.MLPACKAGE_FILES,
-            runtimes: ["python"]
-          },
-          openvino: {
-            url: (0, types_1.hfModelUrl)(HF_REPO, "objectDetection/yolov8/openvino/camstack-yolov8s.xml"),
-            sizeMB: 22,
-            runtimes: ["python"]
-          },
-          tflite: {
-            url: (0, types_1.hfModelUrl)(HF_REPO, "objectDetection/yolov8/tflite/camstack-yolov8s_float32.tflite"),
-            sizeMB: 43,
-            runtimes: ["python"]
-          }
-        }
+      coreml: {
+        url: (0, import_types.hfModelUrl)(HF_REPO, "objectDetection/yolov8/coreml/camstack-yolov8n.mlpackage"),
+        sizeMB: 6,
+        isDirectory: true,
+        files: MLPACKAGE_FILES,
+        runtimes: ["python"]
       },
-      {
-        id: "yolov8s-relu",
-        name: "YOLOv8 Small ReLU",
-        description: "YOLOv8 Small with ReLU activation \u2014 better hardware compatibility",
-        inputSize: { width: 640, height: 640 },
-        labels: types_1.COCO_80_LABELS,
-        formats: {
-          onnx: {
-            url: (0, types_1.hfModelUrl)(HF_REPO, "objectDetection/yolov8/onnx/camstack-yolov8s-relu.onnx"),
-            sizeMB: 43
-          }
-        }
+      openvino: {
+        url: (0, import_types.hfModelUrl)(HF_REPO, "objectDetection/yolov8/openvino/camstack-yolov8n.xml"),
+        sizeMB: 7,
+        runtimes: ["python"]
       },
-      {
-        id: "yolov8m",
-        name: "YOLOv8 Medium",
-        description: "YOLOv8 Medium \u2014 higher accuracy, moderate size",
-        inputSize: { width: 640, height: 640 },
-        labels: types_1.COCO_80_LABELS,
-        formats: {
-          onnx: {
-            url: (0, types_1.hfModelUrl)(HF_REPO, "objectDetection/yolov8/onnx/camstack-yolov8m.onnx"),
-            sizeMB: 99
-          },
-          coreml: {
-            url: (0, types_1.hfModelUrl)(HF_REPO, "objectDetection/yolov8/coreml/camstack-yolov8m.mlpackage"),
-            sizeMB: 49,
-            isDirectory: true,
-            files: exports2.MLPACKAGE_FILES,
-            runtimes: ["python"]
-          },
-          openvino: {
-            url: (0, types_1.hfModelUrl)(HF_REPO, "objectDetection/yolov8/openvino/camstack-yolov8m.xml"),
-            sizeMB: 50,
-            runtimes: ["python"]
-          },
-          tflite: {
-            url: (0, types_1.hfModelUrl)(HF_REPO, "objectDetection/yolov8/tflite/camstack-yolov8m_float32.tflite"),
-            sizeMB: 99,
-            runtimes: ["python"]
-          }
-        }
+      tflite: {
+        url: (0, import_types.hfModelUrl)(HF_REPO, "objectDetection/yolov8/tflite/camstack-yolov8n_float32.tflite"),
+        sizeMB: 12,
+        runtimes: ["python"]
+      }
+    }
+  },
+  {
+    id: "yolov8s",
+    name: "YOLOv8 Small",
+    description: "YOLOv8 Small \u2014 balanced speed and accuracy",
+    inputSize: { width: 640, height: 640 },
+    labels: import_types.COCO_80_LABELS,
+    formats: {
+      onnx: {
+        url: (0, import_types.hfModelUrl)(HF_REPO, "objectDetection/yolov8/onnx/camstack-yolov8s.onnx"),
+        sizeMB: 43
       },
-      {
-        id: "yolov8l",
-        name: "YOLOv8 Large",
-        description: "YOLOv8 Large \u2014 high-accuracy large model",
-        inputSize: { width: 640, height: 640 },
-        labels: types_1.COCO_80_LABELS,
-        formats: {
-          onnx: {
-            url: (0, types_1.hfModelUrl)(HF_REPO, "objectDetection/yolov8/onnx/camstack-yolov8l.onnx"),
-            sizeMB: 167
-          },
-          coreml: {
-            url: (0, types_1.hfModelUrl)(HF_REPO, "objectDetection/yolov8/coreml/camstack-yolov8l.mlpackage"),
-            sizeMB: 83,
-            isDirectory: true,
-            files: exports2.MLPACKAGE_FILES,
-            runtimes: ["python"]
-          },
-          openvino: {
-            url: (0, types_1.hfModelUrl)(HF_REPO, "objectDetection/yolov8/openvino/camstack-yolov8l.xml"),
-            sizeMB: 84,
-            runtimes: ["python"]
-          }
-        }
+      coreml: {
+        url: (0, import_types.hfModelUrl)(HF_REPO, "objectDetection/yolov8/coreml/camstack-yolov8s.mlpackage"),
+        sizeMB: 21,
+        isDirectory: true,
+        files: MLPACKAGE_FILES,
+        runtimes: ["python"]
       },
-      {
-        id: "yolov8x",
-        name: "YOLOv8 Extra-Large",
-        description: "YOLOv8 Extra-Large \u2014 maximum accuracy",
-        inputSize: { width: 640, height: 640 },
-        labels: types_1.COCO_80_LABELS,
-        formats: {
-          onnx: {
-            url: (0, types_1.hfModelUrl)(HF_REPO, "objectDetection/yolov8/onnx/camstack-yolov8x.onnx"),
-            sizeMB: 260
-          },
-          coreml: {
-            url: (0, types_1.hfModelUrl)(HF_REPO, "objectDetection/yolov8/coreml/camstack-yolov8x.mlpackage"),
-            sizeMB: 130,
-            isDirectory: true,
-            files: exports2.MLPACKAGE_FILES,
-            runtimes: ["python"]
-          },
-          openvino: {
-            url: (0, types_1.hfModelUrl)(HF_REPO, "objectDetection/yolov8/openvino/camstack-yolov8x.xml"),
-            sizeMB: 131,
-            runtimes: ["python"]
-          }
-        }
+      openvino: {
+        url: (0, import_types.hfModelUrl)(HF_REPO, "objectDetection/yolov8/openvino/camstack-yolov8s.xml"),
+        sizeMB: 22,
+        runtimes: ["python"]
       },
-      // ── YOLOv9 ──────────────────────────────────────────────────────
-      {
-        id: "yolov9t",
-        name: "YOLOv9 Tiny",
-        description: "YOLOv9 Tiny \u2014 ultra-lightweight next-gen detector",
-        inputSize: { width: 640, height: 640 },
-        labels: types_1.COCO_80_LABELS,
-        formats: {
-          onnx: {
-            url: (0, types_1.hfModelUrl)(HF_REPO, "objectDetection/yolov9/onnx/camstack-yolov9t.onnx"),
-            sizeMB: 8
-          },
-          coreml: {
-            url: (0, types_1.hfModelUrl)(HF_REPO, "objectDetection/yolov9/coreml/camstack-yolov9t.mlpackage"),
-            sizeMB: 4,
-            isDirectory: true,
-            files: exports2.MLPACKAGE_FILES,
-            runtimes: ["python"]
-          },
-          openvino: {
-            url: (0, types_1.hfModelUrl)(HF_REPO, "objectDetection/yolov9/openvino/camstack-yolov9t.xml"),
-            sizeMB: 6,
-            runtimes: ["python"]
-          },
-          tflite: {
-            url: (0, types_1.hfModelUrl)(HF_REPO, "objectDetection/yolov9/tflite/camstack-yolov9t_float32.tflite"),
-            sizeMB: 8,
-            runtimes: ["python"]
-          }
-        }
+      tflite: {
+        url: (0, import_types.hfModelUrl)(HF_REPO, "objectDetection/yolov8/tflite/camstack-yolov8s_float32.tflite"),
+        sizeMB: 43,
+        runtimes: ["python"]
+      }
+    }
+  },
+  {
+    id: "yolov8s-relu",
+    name: "YOLOv8 Small ReLU",
+    description: "YOLOv8 Small with ReLU activation \u2014 better hardware compatibility",
+    inputSize: { width: 640, height: 640 },
+    labels: import_types.COCO_80_LABELS,
+    formats: {
+      onnx: {
+        url: (0, import_types.hfModelUrl)(HF_REPO, "objectDetection/yolov8/onnx/camstack-yolov8s-relu.onnx"),
+        sizeMB: 43
+      }
+    }
+  },
+  {
+    id: "yolov8m",
+    name: "YOLOv8 Medium",
+    description: "YOLOv8 Medium \u2014 higher accuracy, moderate size",
+    inputSize: { width: 640, height: 640 },
+    labels: import_types.COCO_80_LABELS,
+    formats: {
+      onnx: {
+        url: (0, import_types.hfModelUrl)(HF_REPO, "objectDetection/yolov8/onnx/camstack-yolov8m.onnx"),
+        sizeMB: 99
       },
-      {
-        id: "yolov9s",
-        name: "YOLOv9 Small",
-        description: "YOLOv9 Small \u2014 improved efficiency over YOLOv8s",
-        inputSize: { width: 640, height: 640 },
-        labels: types_1.COCO_80_LABELS,
-        formats: {
-          onnx: {
-            url: (0, types_1.hfModelUrl)(HF_REPO, "objectDetection/yolov9/onnx/camstack-yolov9s.onnx"),
-            sizeMB: 28
-          },
-          coreml: {
-            url: (0, types_1.hfModelUrl)(HF_REPO, "objectDetection/yolov9/coreml/camstack-yolov9s.mlpackage"),
-            sizeMB: 14,
-            isDirectory: true,
-            files: exports2.MLPACKAGE_FILES,
-            runtimes: ["python"]
-          },
-          openvino: {
-            url: (0, types_1.hfModelUrl)(HF_REPO, "objectDetection/yolov9/openvino/camstack-yolov9s.xml"),
-            sizeMB: 16,
-            runtimes: ["python"]
-          },
-          tflite: {
-            url: (0, types_1.hfModelUrl)(HF_REPO, "objectDetection/yolov9/tflite/camstack-yolov9s_float32.tflite"),
-            sizeMB: 28,
-            runtimes: ["python"]
-          }
-        }
+      coreml: {
+        url: (0, import_types.hfModelUrl)(HF_REPO, "objectDetection/yolov8/coreml/camstack-yolov8m.mlpackage"),
+        sizeMB: 49,
+        isDirectory: true,
+        files: MLPACKAGE_FILES,
+        runtimes: ["python"]
       },
-      {
-        id: "yolov9c",
-        name: "YOLOv9 C",
-        description: "YOLOv9 C \u2014 high-accuracy compact model",
-        inputSize: { width: 640, height: 640 },
-        labels: types_1.COCO_80_LABELS,
-        formats: {
-          onnx: {
-            url: (0, types_1.hfModelUrl)(HF_REPO, "objectDetection/yolov9/onnx/camstack-yolov9c.onnx"),
-            sizeMB: 97
-          },
-          coreml: {
-            url: (0, types_1.hfModelUrl)(HF_REPO, "objectDetection/yolov9/coreml/camstack-yolov9c.mlpackage"),
-            sizeMB: 48,
-            isDirectory: true,
-            files: exports2.MLPACKAGE_FILES,
-            runtimes: ["python"]
-          },
-          openvino: {
-            url: (0, types_1.hfModelUrl)(HF_REPO, "objectDetection/yolov9/openvino/camstack-yolov9c.xml"),
-            sizeMB: 49,
-            runtimes: ["python"]
-          },
-          tflite: {
-            url: (0, types_1.hfModelUrl)(HF_REPO, "objectDetection/yolov9/tflite/camstack-yolov9c_float32.tflite"),
-            sizeMB: 97,
-            runtimes: ["python"]
-          }
-        }
+      openvino: {
+        url: (0, import_types.hfModelUrl)(HF_REPO, "objectDetection/yolov8/openvino/camstack-yolov8m.xml"),
+        sizeMB: 50,
+        runtimes: ["python"]
       },
-      // ── YOLO11 ────────────────────────────────────────────────────
-      {
-        id: "yolo11n",
-        name: "YOLO11 Nano",
-        description: "YOLO11 Nano \u2014 fastest, smallest YOLO11 detection model (mAP 39.5)",
-        inputSize: { width: 640, height: 640 },
-        labels: types_1.COCO_80_LABELS,
-        formats: {
-          onnx: {
-            url: (0, types_1.hfModelUrl)(HF_REPO, "objectDetection/yolo11/onnx/camstack-yolo11n.onnx"),
-            sizeMB: 10
-          },
-          coreml: {
-            url: (0, types_1.hfModelUrl)(HF_REPO, "objectDetection/yolo11/coreml/camstack-yolo11n.mlpackage"),
-            sizeMB: 5,
-            isDirectory: true,
-            files: exports2.MLPACKAGE_FILES,
-            runtimes: ["python"]
-          },
-          openvino: {
-            url: (0, types_1.hfModelUrl)(HF_REPO, "objectDetection/yolo11/openvino/camstack-yolo11n.xml"),
-            sizeMB: 5,
-            runtimes: ["python"]
-          },
-          tflite: {
-            url: (0, types_1.hfModelUrl)(HF_REPO, "objectDetection/yolo11/tflite/camstack-yolo11n_float32.tflite"),
-            sizeMB: 10,
-            runtimes: ["python"]
-          }
-        }
+      tflite: {
+        url: (0, import_types.hfModelUrl)(HF_REPO, "objectDetection/yolov8/tflite/camstack-yolov8m_float32.tflite"),
+        sizeMB: 99,
+        runtimes: ["python"]
+      }
+    }
+  },
+  {
+    id: "yolov8l",
+    name: "YOLOv8 Large",
+    description: "YOLOv8 Large \u2014 high-accuracy large model",
+    inputSize: { width: 640, height: 640 },
+    labels: import_types.COCO_80_LABELS,
+    formats: {
+      onnx: {
+        url: (0, import_types.hfModelUrl)(HF_REPO, "objectDetection/yolov8/onnx/camstack-yolov8l.onnx"),
+        sizeMB: 167
       },
-      {
-        id: "yolo11s",
-        name: "YOLO11 Small",
-        description: "YOLO11 Small \u2014 balanced speed and accuracy (mAP 47.0)",
-        inputSize: { width: 640, height: 640 },
-        labels: types_1.COCO_80_LABELS,
-        formats: {
-          onnx: {
-            url: (0, types_1.hfModelUrl)(HF_REPO, "objectDetection/yolo11/onnx/camstack-yolo11s.onnx"),
-            sizeMB: 36
-          },
-          coreml: {
-            url: (0, types_1.hfModelUrl)(HF_REPO, "objectDetection/yolo11/coreml/camstack-yolo11s.mlpackage"),
-            sizeMB: 18,
-            isDirectory: true,
-            files: exports2.MLPACKAGE_FILES,
-            runtimes: ["python"]
-          },
-          openvino: {
-            url: (0, types_1.hfModelUrl)(HF_REPO, "objectDetection/yolo11/openvino/camstack-yolo11s.xml"),
-            sizeMB: 18,
-            runtimes: ["python"]
-          },
-          tflite: {
-            url: (0, types_1.hfModelUrl)(HF_REPO, "objectDetection/yolo11/tflite/camstack-yolo11s_float32.tflite"),
-            sizeMB: 36,
-            runtimes: ["python"]
-          }
-        }
+      coreml: {
+        url: (0, import_types.hfModelUrl)(HF_REPO, "objectDetection/yolov8/coreml/camstack-yolov8l.mlpackage"),
+        sizeMB: 83,
+        isDirectory: true,
+        files: MLPACKAGE_FILES,
+        runtimes: ["python"]
       },
-      {
-        id: "yolo11m",
-        name: "YOLO11 Medium",
-        description: "YOLO11 Medium \u2014 higher accuracy, moderate size (mAP 51.5)",
-        inputSize: { width: 640, height: 640 },
-        labels: types_1.COCO_80_LABELS,
-        formats: {
-          onnx: {
-            url: (0, types_1.hfModelUrl)(HF_REPO, "objectDetection/yolo11/onnx/camstack-yolo11m.onnx"),
-            sizeMB: 77
-          },
-          coreml: {
-            url: (0, types_1.hfModelUrl)(HF_REPO, "objectDetection/yolo11/coreml/camstack-yolo11m.mlpackage"),
-            sizeMB: 39,
-            isDirectory: true,
-            files: exports2.MLPACKAGE_FILES,
-            runtimes: ["python"]
-          },
-          openvino: {
-            url: (0, types_1.hfModelUrl)(HF_REPO, "objectDetection/yolo11/openvino/camstack-yolo11m.xml"),
-            sizeMB: 39,
-            runtimes: ["python"]
-          },
-          tflite: {
-            url: (0, types_1.hfModelUrl)(HF_REPO, "objectDetection/yolo11/tflite/camstack-yolo11m_float32.tflite"),
-            sizeMB: 77,
-            runtimes: ["python"]
-          }
-        }
+      openvino: {
+        url: (0, import_types.hfModelUrl)(HF_REPO, "objectDetection/yolov8/openvino/camstack-yolov8l.xml"),
+        sizeMB: 84,
+        runtimes: ["python"]
+      }
+    }
+  },
+  {
+    id: "yolov8x",
+    name: "YOLOv8 Extra-Large",
+    description: "YOLOv8 Extra-Large \u2014 maximum accuracy",
+    inputSize: { width: 640, height: 640 },
+    labels: import_types.COCO_80_LABELS,
+    formats: {
+      onnx: {
+        url: (0, import_types.hfModelUrl)(HF_REPO, "objectDetection/yolov8/onnx/camstack-yolov8x.onnx"),
+        sizeMB: 260
       },
-      {
-        id: "yolo11l",
-        name: "YOLO11 Large",
-        description: "YOLO11 Large \u2014 high-accuracy large model (mAP 53.4)",
-        inputSize: { width: 640, height: 640 },
-        labels: types_1.COCO_80_LABELS,
-        formats: {
-          onnx: {
-            url: (0, types_1.hfModelUrl)(HF_REPO, "objectDetection/yolo11/onnx/camstack-yolo11l.onnx"),
-            sizeMB: 97
-          },
-          coreml: {
-            url: (0, types_1.hfModelUrl)(HF_REPO, "objectDetection/yolo11/coreml/camstack-yolo11l.mlpackage"),
-            sizeMB: 49,
-            isDirectory: true,
-            files: exports2.MLPACKAGE_FILES,
-            runtimes: ["python"]
-          },
-          openvino: {
-            url: (0, types_1.hfModelUrl)(HF_REPO, "objectDetection/yolo11/openvino/camstack-yolo11l.xml"),
-            sizeMB: 49,
-            runtimes: ["python"]
-          },
-          tflite: {
-            url: (0, types_1.hfModelUrl)(HF_REPO, "objectDetection/yolo11/tflite/camstack-yolo11l_float32.tflite"),
-            sizeMB: 97,
-            runtimes: ["python"]
-          }
-        }
+      coreml: {
+        url: (0, import_types.hfModelUrl)(HF_REPO, "objectDetection/yolov8/coreml/camstack-yolov8x.mlpackage"),
+        sizeMB: 130,
+        isDirectory: true,
+        files: MLPACKAGE_FILES,
+        runtimes: ["python"]
       },
-      {
-        id: "yolo11x",
-        name: "YOLO11 Extra-Large",
-        description: "YOLO11 Extra-Large \u2014 maximum accuracy (mAP 54.7)",
-        inputSize: { width: 640, height: 640 },
-        labels: types_1.COCO_80_LABELS,
-        formats: {
-          onnx: {
-            url: (0, types_1.hfModelUrl)(HF_REPO, "objectDetection/yolo11/onnx/camstack-yolo11x.onnx"),
-            sizeMB: 218
-          },
-          coreml: {
-            url: (0, types_1.hfModelUrl)(HF_REPO, "objectDetection/yolo11/coreml/camstack-yolo11x.mlpackage"),
-            sizeMB: 109,
-            isDirectory: true,
-            files: exports2.MLPACKAGE_FILES,
-            runtimes: ["python"]
-          },
-          openvino: {
-            url: (0, types_1.hfModelUrl)(HF_REPO, "objectDetection/yolo11/openvino/camstack-yolo11x.xml"),
-            sizeMB: 109,
-            runtimes: ["python"]
-          },
-          tflite: {
-            url: (0, types_1.hfModelUrl)(HF_REPO, "objectDetection/yolo11/tflite/camstack-yolo11x_float32.tflite"),
-            sizeMB: 218,
-            runtimes: ["python"]
-          }
-        }
+      openvino: {
+        url: (0, import_types.hfModelUrl)(HF_REPO, "objectDetection/yolov8/openvino/camstack-yolov8x.xml"),
+        sizeMB: 131,
+        runtimes: ["python"]
       }
-    ];
-  }
-});
-
-// src/catalogs/segmentation-models.js
-var require_segmentation_models = __commonJS({
-  "src/catalogs/segmentation-models.js"(exports2) {
-    "use strict";
-    Object.defineProperty(exports2, "__esModule", { value: true });
-    exports2.SEGMENTATION_MODELS = void 0;
-    var types_1 = require("@camstack/types");
-    var object_detection_models_js_1 = require_object_detection_models();
-    var HF_REPO = "camstack/camstack-models";
-    exports2.SEGMENTATION_MODELS = [
-      // ── YOLO11-seg ───────────────────────────────────────────────
-      {
-        id: "yolo11n-seg",
-        name: "YOLO11 Nano Segmentation",
-        description: "YOLO11 Nano \u2014 fastest, smallest YOLO11 instance segmentation model",
-        inputSize: { width: 640, height: 640 },
-        labels: types_1.COCO_80_LABELS,
-        formats: {
-          onnx: {
-            url: (0, types_1.hfModelUrl)(HF_REPO, "segmentation/yolo11-seg/onnx/camstack-yolo11n-seg.onnx"),
-            sizeMB: 11
-          },
-          coreml: {
-            url: (0, types_1.hfModelUrl)(HF_REPO, "segmentation/yolo11-seg/coreml/camstack-yolo11n-seg.mlpackage"),
-            sizeMB: 6,
-            isDirectory: true,
-            files: object_detection_models_js_1.MLPACKAGE_FILES,
-            runtimes: ["python"]
-          },
-          openvino: {
-            url: (0, types_1.hfModelUrl)(HF_REPO, "segmentation/yolo11-seg/openvino/camstack-yolo11n-seg.xml"),
-            sizeMB: 6,
-            runtimes: ["python"]
-          }
-        }
+    }
+  },
+  // ── YOLOv9 ──────────────────────────────────────────────────────
+  {
+    id: "yolov9t",
+    name: "YOLOv9 Tiny",
+    description: "YOLOv9 Tiny \u2014 ultra-lightweight next-gen detector",
+    inputSize: { width: 640, height: 640 },
+    labels: import_types.COCO_80_LABELS,
+    formats: {
+      onnx: {
+        url: (0, import_types.hfModelUrl)(HF_REPO, "objectDetection/yolov9/onnx/camstack-yolov9t.onnx"),
+        sizeMB: 8
       },
-      {
-        id: "yolo11s-seg",
-        name: "YOLO11 Small Segmentation",
-        description: "YOLO11 Small \u2014 balanced speed and accuracy for instance segmentation",
-        inputSize: { width: 640, height: 640 },
-        labels: types_1.COCO_80_LABELS,
-        formats: {
-          onnx: {
-            url: (0, types_1.hfModelUrl)(HF_REPO, "segmentation/yolo11-seg/onnx/camstack-yolo11s-seg.onnx"),
-            sizeMB: 39
-          },
-          coreml: {
-            url: (0, types_1.hfModelUrl)(HF_REPO, "segmentation/yolo11-seg/coreml/camstack-yolo11s-seg.mlpackage"),
-            sizeMB: 20,
-            isDirectory: true,
-            files: object_detection_models_js_1.MLPACKAGE_FILES,
-            runtimes: ["python"]
-          },
-          openvino: {
-            url: (0, types_1.hfModelUrl)(HF_REPO, "segmentation/yolo11-seg/openvino/camstack-yolo11s-seg.xml"),
-            sizeMB: 20,
-            runtimes: ["python"]
-          }
-        }
+      coreml: {
+        url: (0, import_types.hfModelUrl)(HF_REPO, "objectDetection/yolov9/coreml/camstack-yolov9t.mlpackage"),
+        sizeMB: 4,
+        isDirectory: true,
+        files: MLPACKAGE_FILES,
+        runtimes: ["python"]
       },
-      {
-        id: "yolo11m-seg",
-        name: "YOLO11 Medium Segmentation",
-        description: "YOLO11 Medium \u2014 higher accuracy instance segmentation",
-        inputSize: { width: 640, height: 640 },
-        labels: types_1.COCO_80_LABELS,
-        formats: {
-          onnx: {
-            url: (0, types_1.hfModelUrl)(HF_REPO, "segmentation/yolo11-seg/onnx/camstack-yolo11m-seg.onnx"),
-            sizeMB: 86
-          },
-          coreml: {
-            url: (0, types_1.hfModelUrl)(HF_REPO, "segmentation/yolo11-seg/coreml/camstack-yolo11m-seg.mlpackage"),
-            sizeMB: 43,
-            isDirectory: true,
-            files: object_detection_models_js_1.MLPACKAGE_FILES,
-            runtimes: ["python"]
-          },
-          openvino: {
-            url: (0, types_1.hfModelUrl)(HF_REPO, "segmentation/yolo11-seg/openvino/camstack-yolo11m-seg.xml"),
-            sizeMB: 43,
-            runtimes: ["python"]
-          }
-        }
+      openvino: {
+        url: (0, import_types.hfModelUrl)(HF_REPO, "objectDetection/yolov9/openvino/camstack-yolov9t.xml"),
+        sizeMB: 6,
+        runtimes: ["python"]
       },
-      // ── YOLOv8-seg — CoreML supported ─────────────────────────────
-      {
-        id: "yolov8n-seg",
-        name: "YOLOv8 Nano Segmentation",
-        description: "YOLOv8 Nano \u2014 fastest, smallest YOLOv8 instance segmentation model",
-        inputSize: { width: 640, height: 640 },
-        labels: types_1.COCO_80_LABELS,
-        formats: {
-          onnx: {
-            url: (0, types_1.hfModelUrl)(HF_REPO, "segmentation/yolov8-seg/onnx/camstack-yolov8n-seg.onnx"),
-            sizeMB: 13
-          },
-          coreml: {
-            url: (0, types_1.hfModelUrl)(HF_REPO, "segmentation/yolov8-seg/coreml/camstack-yolov8n-seg.mlpackage"),
-            sizeMB: 7,
-            isDirectory: true,
-            files: object_detection_models_js_1.MLPACKAGE_FILES,
-            runtimes: ["python"]
-          },
-          openvino: {
-            url: (0, types_1.hfModelUrl)(HF_REPO, "segmentation/yolov8-seg/openvino/camstack-yolov8n-seg.xml"),
-            sizeMB: 7,
-            runtimes: ["python"]
-          }
-        }
+      tflite: {
+        url: (0, import_types.hfModelUrl)(HF_REPO, "objectDetection/yolov9/tflite/camstack-yolov9t_float32.tflite"),
+        sizeMB: 8,
+        runtimes: ["python"]
+      }
+    }
+  },
+  {
+    id: "yolov9s",
+    name: "YOLOv9 Small",
+    description: "YOLOv9 Small \u2014 improved efficiency over YOLOv8s",
+    inputSize: { width: 640, height: 640 },
+    labels: import_types.COCO_80_LABELS,
+    formats: {
+      onnx: {
+        url: (0, import_types.hfModelUrl)(HF_REPO, "objectDetection/yolov9/onnx/camstack-yolov9s.onnx"),
+        sizeMB: 28
       },
-      {
-        id: "yolov8s-seg",
-        name: "YOLOv8 Small Segmentation",
-        description: "YOLOv8 Small \u2014 balanced speed and accuracy for instance segmentation",
-        inputSize: { width: 640, height: 640 },
-        labels: types_1.COCO_80_LABELS,
-        formats: {
-          onnx: {
-            url: (0, types_1.hfModelUrl)(HF_REPO, "segmentation/yolov8-seg/onnx/camstack-yolov8s-seg.onnx"),
-            sizeMB: 45
-          },
-          coreml: {
-            url: (0, types_1.hfModelUrl)(HF_REPO, "segmentation/yolov8-seg/coreml/camstack-yolov8s-seg.mlpackage"),
-            sizeMB: 23,
-            isDirectory: true,
-            files: object_detection_models_js_1.MLPACKAGE_FILES,
-            runtimes: ["python"]
-          },
-          openvino: {
-            url: (0, types_1.hfModelUrl)(HF_REPO, "segmentation/yolov8-seg/openvino/camstack-yolov8s-seg.xml"),
-            sizeMB: 23,
-            runtimes: ["python"]
-          }
-        }
+      coreml: {
+        url: (0, import_types.hfModelUrl)(HF_REPO, "objectDetection/yolov9/coreml/camstack-yolov9s.mlpackage"),
+        sizeMB: 14,
+        isDirectory: true,
+        files: MLPACKAGE_FILES,
+        runtimes: ["python"]
       },
-      {
-        id: "yolov8m-seg",
-        name: "YOLOv8 Medium Segmentation",
-        description: "YOLOv8 Medium \u2014 higher accuracy instance segmentation",
-        inputSize: { width: 640, height: 640 },
-        labels: types_1.COCO_80_LABELS,
-        formats: {
-          onnx: {
-            url: (0, types_1.hfModelUrl)(HF_REPO, "segmentation/yolov8-seg/onnx/camstack-yolov8m-seg.onnx"),
-            sizeMB: 104
-          },
-          coreml: {
-            url: (0, types_1.hfModelUrl)(HF_REPO, "segmentation/yolov8-seg/coreml/camstack-yolov8m-seg.mlpackage"),
-            sizeMB: 52,
-            isDirectory: true,
-            files: object_detection_models_js_1.MLPACKAGE_FILES,
-            runtimes: ["python"]
-          },
-          openvino: {
-            url: (0, types_1.hfModelUrl)(HF_REPO, "segmentation/yolov8-seg/openvino/camstack-yolov8m-seg.xml"),
-            sizeMB: 53,
-            runtimes: ["python"]
-          }
-        }
+      openvino: {
+        url: (0, import_types.hfModelUrl)(HF_REPO, "objectDetection/yolov9/openvino/camstack-yolov9s.xml"),
+        sizeMB: 16,
+        runtimes: ["python"]
+      },
+      tflite: {
+        url: (0, import_types.hfModelUrl)(HF_REPO, "objectDetection/yolov9/tflite/camstack-yolov9s_float32.tflite"),
+        sizeMB: 28,
+        runtimes: ["python"]
       }
-    ];
-  }
-});
-
-// src/shared/image-utils.js
-var require_image_utils = __commonJS({
-  "src/shared/image-utils.js"(exports2) {
-    "use strict";
-    var __importDefault = exports2 && exports2.__importDefault || function(mod) {
-      return mod && mod.__esModule ? mod : { "default": mod };
-    };
-    Object.defineProperty(exports2, "__esModule", { value: true });
-    exports2.jpegToRgb = jpegToRgb;
-    exports2.cropRegion = cropRegion;
-    exports2.letterbox = letterbox2;
-    exports2.resizeAndNormalize = resizeAndNormalize;
-    exports2.rgbToGrayscale = rgbToGrayscale;
-    var sharp_1 = __importDefault(require("sharp"));
-    async function jpegToRgb(jpeg) {
-      const { data, info } = await (0, sharp_1.default)(jpeg).removeAlpha().raw().toBuffer({ resolveWithObject: true });
-      return { data, width: info.width, height: info.height };
     }
-    async function cropRegion(jpeg, roi) {
-      return (0, sharp_1.default)(jpeg).extract({
-        left: Math.round(roi.x),
-        top: Math.round(roi.y),
-        width: Math.round(roi.w),
-        height: Math.round(roi.h)
-      }).jpeg().toBuffer();
-    }
-    async function letterbox2(jpeg, targetSize) {
-      const meta = await (0, sharp_1.default)(jpeg).metadata();
-      const originalWidth = meta.width ?? 0;
-      const originalHeight = meta.height ?? 0;
-      const scale = Math.min(targetSize / originalWidth, targetSize / originalHeight);
-      const scaledWidth = Math.round(originalWidth * scale);
-      const scaledHeight = Math.round(originalHeight * scale);
-      const padX = Math.floor((targetSize - scaledWidth) / 2);
-      const padY = Math.floor((targetSize - scaledHeight) / 2);
-      const { data } = await (0, sharp_1.default)(jpeg).resize(scaledWidth, scaledHeight).extend({
-        top: padY,
-        bottom: targetSize - scaledHeight - padY,
-        left: padX,
-        right: targetSize - scaledWidth - padX,
-        background: { r: 114, g: 114, b: 114 }
-      }).removeAlpha().raw().toBuffer({ resolveWithObject: true });
-      const numPixels = targetSize * targetSize;
-      const float32 = new Float32Array(3 * numPixels);
-      for (let i = 0; i < numPixels; i++) {
-        const srcBase = i * 3;
-        float32[0 * numPixels + i] = data[srcBase] / 255;
-        float32[1 * numPixels + i] = data[srcBase + 1] / 255;
-        float32[2 * numPixels + i] = data[srcBase + 2] / 255;
+  },
+  {
+    id: "yolov9c",
+    name: "YOLOv9 C",
+    description: "YOLOv9 C \u2014 high-accuracy compact model",
+    inputSize: { width: 640, height: 640 },
+    labels: import_types.COCO_80_LABELS,
+    formats: {
+      onnx: {
+        url: (0, import_types.hfModelUrl)(HF_REPO, "objectDetection/yolov9/onnx/camstack-yolov9c.onnx"),
+        sizeMB: 97
+      },
+      coreml: {
+        url: (0, import_types.hfModelUrl)(HF_REPO, "objectDetection/yolov9/coreml/camstack-yolov9c.mlpackage"),
+        sizeMB: 48,
+        isDirectory: true,
+        files: MLPACKAGE_FILES,
+        runtimes: ["python"]
+      },
+      openvino: {
+        url: (0, import_types.hfModelUrl)(HF_REPO, "objectDetection/yolov9/openvino/camstack-yolov9c.xml"),
+        sizeMB: 49,
+        runtimes: ["python"]
+      },
+      tflite: {
+        url: (0, import_types.hfModelUrl)(HF_REPO, "objectDetection/yolov9/tflite/camstack-yolov9c_float32.tflite"),
+        sizeMB: 97,
+        runtimes: ["python"]
       }
-      return { data: float32, scale, padX, padY, originalWidth, originalHeight };
     }
-    async function resizeAndNormalize(jpeg, targetWidth, targetHeight, normalization, layout) {
-      const { data } = await (0, sharp_1.default)(jpeg).resize(targetWidth, targetHeight, { fit: "fill" }).removeAlpha().raw().toBuffer({ resolveWithObject: true });
-      const numPixels = targetWidth * targetHeight;
-      const float32 = new Float32Array(3 * numPixels);
-      const mean = [0.485, 0.456, 0.406];
-      const std = [0.229, 0.224, 0.225];
-      if (layout === "nchw") {
-        for (let i = 0; i < numPixels; i++) {
-          const srcBase = i * 3;
-          for (let c = 0; c < 3; c++) {
-            const raw = data[srcBase + c] / 255;
-            let val;
-            if (normalization === "zero-one") {
-              val = raw;
-            } else if (normalization === "imagenet") {
-              val = (raw - mean[c]) / std[c];
-            } else {
-              val = data[srcBase + c];
-            }
-            float32[c * numPixels + i] = val;
-          }
-        }
-      } else {
-        for (let i = 0; i < numPixels; i++) {
-          const srcBase = i * 3;
-          for (let c = 0; c < 3; c++) {
-            const raw = data[srcBase + c] / 255;
-            let val;
-            if (normalization === "zero-one") {
-              val = raw;
-            } else if (normalization === "imagenet") {
-              val = (raw - mean[c]) / std[c];
-            } else {
-              val = data[srcBase + c];
-            }
-            float32[i * 3 + c] = val;
-          }
-        }
+  },
+  // ── YOLO11 ────────────────────────────────────────────────────
+  {
+    id: "yolo11n",
+    name: "YOLO11 Nano",
+    description: "YOLO11 Nano \u2014 fastest, smallest YOLO11 detection model (mAP 39.5)",
+    inputSize: { width: 640, height: 640 },
+    labels: import_types.COCO_80_LABELS,
+    formats: {
+      onnx: {
+        url: (0, import_types.hfModelUrl)(HF_REPO, "objectDetection/yolo11/onnx/camstack-yolo11n.onnx"),
+        sizeMB: 10
+      },
+      coreml: {
+        url: (0, import_types.hfModelUrl)(HF_REPO, "objectDetection/yolo11/coreml/camstack-yolo11n.mlpackage"),
+        sizeMB: 5,
+        isDirectory: true,
+        files: MLPACKAGE_FILES,
+        runtimes: ["python"]
+      },
+      openvino: {
+        url: (0, import_types.hfModelUrl)(HF_REPO, "objectDetection/yolo11/openvino/camstack-yolo11n.xml"),
+        sizeMB: 5,
+        runtimes: ["python"]
+      },
+      tflite: {
+        url: (0, import_types.hfModelUrl)(HF_REPO, "objectDetection/yolo11/tflite/camstack-yolo11n_float32.tflite"),
+        sizeMB: 10,
+        runtimes: ["python"]
       }
-      return float32;
     }
-    function rgbToGrayscale(rgb, width, height) {
-      const numPixels = width * height;
-      const gray = new Uint8Array(numPixels);
-      for (let i = 0; i < numPixels; i++) {
-        const r = rgb[i * 3];
-        const g = rgb[i * 3 + 1];
-        const b = rgb[i * 3 + 2];
-        gray[i] = Math.round(0.299 * r + 0.587 * g + 0.114 * b);
+  },
+  {
+    id: "yolo11s",
+    name: "YOLO11 Small",
+    description: "YOLO11 Small \u2014 balanced speed and accuracy (mAP 47.0)",
+    inputSize: { width: 640, height: 640 },
+    labels: import_types.COCO_80_LABELS,
+    formats: {
+      onnx: {
+        url: (0, import_types.hfModelUrl)(HF_REPO, "objectDetection/yolo11/onnx/camstack-yolo11s.onnx"),
+        sizeMB: 36
+      },
+      coreml: {
+        url: (0, import_types.hfModelUrl)(HF_REPO, "objectDetection/yolo11/coreml/camstack-yolo11s.mlpackage"),
+        sizeMB: 18,
+        isDirectory: true,
+        files: MLPACKAGE_FILES,
+        runtimes: ["python"]
+      },
+      openvino: {
+        url: (0, import_types.hfModelUrl)(HF_REPO, "objectDetection/yolo11/openvino/camstack-yolo11s.xml"),
+        sizeMB: 18,
+        runtimes: ["python"]
+      },
+      tflite: {
+        url: (0, import_types.hfModelUrl)(HF_REPO, "objectDetection/yolo11/tflite/camstack-yolo11s_float32.tflite"),
+        sizeMB: 36,
+        runtimes: ["python"]
       }
-      return gray;
     }
-  }
-});
-
-// src/shared/postprocess/yolo.js
-var require_yolo = __commonJS({
-  "src/shared/postprocess/yolo.js"(exports2) {
-    "use strict";
-    Object.defineProperty(exports2, "__esModule", { value: true });
-    exports2.iou = iou;
-    exports2.nms = nms;
-    exports2.yoloPostprocess = yoloPostprocess2;
-    function iou(a, b) {
-      const ax1 = a.x;
-      const ay1 = a.y;
-      const ax2 = a.x + a.w;
-      const ay2 = a.y + a.h;
-      const bx1 = b.x;
-      const by1 = b.y;
-      const bx2 = b.x + b.w;
-      const by2 = b.y + b.h;
-      const interX1 = Math.max(ax1, bx1);
-      const interY1 = Math.max(ay1, by1);
-      const interX2 = Math.min(ax2, bx2);
-      const interY2 = Math.min(ay2, by2);
-      const interW = Math.max(0, interX2 - interX1);
-      const interH = Math.max(0, interY2 - interY1);
-      const interArea = interW * interH;
-      if (interArea === 0)
-        return 0;
-      const areaA = a.w * a.h;
-      const areaB = b.w * b.h;
-      const unionArea = areaA + areaB - interArea;
-      return unionArea === 0 ? 0 : interArea / unionArea;
+  },
+  {
+    id: "yolo11m",
+    name: "YOLO11 Medium",
+    description: "YOLO11 Medium \u2014 higher accuracy, moderate size (mAP 51.5)",
+    inputSize: { width: 640, height: 640 },
+    labels: import_types.COCO_80_LABELS,
+    formats: {
+      onnx: {
+        url: (0, import_types.hfModelUrl)(HF_REPO, "objectDetection/yolo11/onnx/camstack-yolo11m.onnx"),
+        sizeMB: 77
+      },
+      coreml: {
+        url: (0, import_types.hfModelUrl)(HF_REPO, "objectDetection/yolo11/coreml/camstack-yolo11m.mlpackage"),
+        sizeMB: 39,
+        isDirectory: true,
+        files: MLPACKAGE_FILES,
+        runtimes: ["python"]
+      },
+      openvino: {
+        url: (0, import_types.hfModelUrl)(HF_REPO, "objectDetection/yolo11/openvino/camstack-yolo11m.xml"),
+        sizeMB: 39,
+        runtimes: ["python"]
+      },
+      tflite: {
+        url: (0, import_types.hfModelUrl)(HF_REPO, "objectDetection/yolo11/tflite/camstack-yolo11m_float32.tflite"),
+        sizeMB: 77,
+        runtimes: ["python"]
+      }
     }
-    function nms(boxes, iouThreshold) {
-      const indices = boxes.map((_, i) => i).sort((a, b) => boxes[b].score - boxes[a].score);
-      const kept = [];
-      const suppressed = /* @__PURE__ */ new Set();
-      for (const idx of indices) {
-        if (suppressed.has(idx))
-          continue;
-        kept.push(idx);
-        for (const other of indices) {
-          if (other === idx || suppressed.has(other))
-            continue;
-          if (iou(boxes[idx].bbox, boxes[other].bbox) > iouThreshold) {
-            suppressed.add(other);
-          }
-        }
+  },
+  {
+    id: "yolo11l",
+    name: "YOLO11 Large",
+    description: "YOLO11 Large \u2014 high-accuracy large model (mAP 53.4)",
+    inputSize: { width: 640, height: 640 },
+    labels: import_types.COCO_80_LABELS,
+    formats: {
+      onnx: {
+        url: (0, import_types.hfModelUrl)(HF_REPO, "objectDetection/yolo11/onnx/camstack-yolo11l.onnx"),
+        sizeMB: 97
+      },
+      coreml: {
+        url: (0, import_types.hfModelUrl)(HF_REPO, "objectDetection/yolo11/coreml/camstack-yolo11l.mlpackage"),
+        sizeMB: 49,
+        isDirectory: true,
+        files: MLPACKAGE_FILES,
+        runtimes: ["python"]
+      },
+      openvino: {
+        url: (0, import_types.hfModelUrl)(HF_REPO, "objectDetection/yolo11/openvino/camstack-yolo11l.xml"),
+        sizeMB: 49,
+        runtimes: ["python"]
+      },
+      tflite: {
+        url: (0, import_types.hfModelUrl)(HF_REPO, "objectDetection/yolo11/tflite/camstack-yolo11l_float32.tflite"),
+        sizeMB: 97,
+        runtimes: ["python"]
       }
-      return kept;
     }
-    function yoloPostprocess2(output, numClasses, numBoxes, options) {
-      const { confidence, iouThreshold, labels, scale, padX, padY, originalWidth, originalHeight } = options;
-      const candidates = [];
-      for (let i = 0; i < numBoxes; i++) {
-        const cx = output[0 * numBoxes + i];
-        const cy = output[1 * numBoxes + i];
-        const w = output[2 * numBoxes + i];
-        const h = output[3 * numBoxes + i];
-        let bestScore = -Infinity;
-        let bestClass = 0;
-        for (let j = 0; j < numClasses; j++) {
-          const score = output[(4 + j) * numBoxes + i];
-          if (score > bestScore) {
-            bestScore = score;
-            bestClass = j;
-          }
-        }
-        if (bestScore < confidence)
-          continue;
-        const bbox = {
-          x: cx - w / 2,
-          y: cy - h / 2,
-          w,
-          h
-        };
-        candidates.push({ bbox, score: bestScore, classIdx: bestClass });
+  },
+  {
+    id: "yolo11x",
+    name: "YOLO11 Extra-Large",
+    description: "YOLO11 Extra-Large \u2014 maximum accuracy (mAP 54.7)",
+    inputSize: { width: 640, height: 640 },
+    labels: import_types.COCO_80_LABELS,
+    formats: {
+      onnx: {
+        url: (0, import_types.hfModelUrl)(HF_REPO, "objectDetection/yolo11/onnx/camstack-yolo11x.onnx"),
+        sizeMB: 218
+      },
+      coreml: {
+        url: (0, import_types.hfModelUrl)(HF_REPO, "objectDetection/yolo11/coreml/camstack-yolo11x.mlpackage"),
+        sizeMB: 109,
+        isDirectory: true,
+        files: MLPACKAGE_FILES,
+        runtimes: ["python"]
+      },
+      openvino: {
+        url: (0, import_types.hfModelUrl)(HF_REPO, "objectDetection/yolo11/openvino/camstack-yolo11x.xml"),
+        sizeMB: 109,
+        runtimes: ["python"]
+      },
+      tflite: {
+        url: (0, import_types.hfModelUrl)(HF_REPO, "objectDetection/yolo11/tflite/camstack-yolo11x_float32.tflite"),
+        sizeMB: 218,
+        runtimes: ["python"]
       }
-      if (candidates.length === 0)
-        return [];
-      const keptIndices = nms(candidates, iouThreshold);
-      return keptIndices.map((idx) => {
-        const { bbox, score, classIdx } = candidates[idx];
-        const label = labels[classIdx] ?? String(classIdx);
-        const x = Math.max(0, Math.min(originalWidth, (bbox.x - padX) / scale));
-        const y = Math.max(0, Math.min(originalHeight, (bbox.y - padY) / scale));
-        const x2 = Math.max(0, Math.min(originalWidth, (bbox.x + bbox.w - padX) / scale));
-        const y2 = Math.max(0, Math.min(originalHeight, (bbox.y + bbox.h - padY) / scale));
-        const finalBbox = { x, y, w: x2 - x, h: y2 - y };
-        return {
-          class: label,
-          originalClass: label,
-          score,
-          bbox: finalBbox
-        };
-      });
     }
   }
-});
+];
 
-// src/shared/postprocess/yolo-seg.js
-var require_yolo_seg = __commonJS({
-  "src/shared/postprocess/yolo-seg.js"(exports2) {
-    "use strict";
-    Object.defineProperty(exports2, "__esModule", { value: true });
-    exports2.sigmoid = sigmoid;
-    exports2.computeRawMask = computeRawMask;
-    exports2.cropAndThresholdMask = cropAndThresholdMask;
-    exports2.yoloSegPostprocess = yoloSegPostprocess2;
-    var yolo_js_1 = require_yolo();
-    function sigmoid(x) {
-      return 1 / (1 + Math.exp(-x));
+// src/catalogs/segmentation-models.ts
+var import_types2 = require("@camstack/types");
+var HF_REPO2 = "camstack/camstack-models";
+var SEGMENTATION_MODELS = [
+  // ── YOLO11-seg ───────────────────────────────────────────────
+  {
+    id: "yolo11n-seg",
+    name: "YOLO11 Nano Segmentation",
+    description: "YOLO11 Nano \u2014 fastest, smallest YOLO11 instance segmentation model",
+    inputSize: { width: 640, height: 640 },
+    labels: import_types2.COCO_80_LABELS,
+    formats: {
+      onnx: {
+        url: (0, import_types2.hfModelUrl)(HF_REPO2, "segmentation/yolo11-seg/onnx/camstack-yolo11n-seg.onnx"),
+        sizeMB: 11
+      },
+      coreml: {
+        url: (0, import_types2.hfModelUrl)(HF_REPO2, "segmentation/yolo11-seg/coreml/camstack-yolo11n-seg.mlpackage"),
+        sizeMB: 6,
+        isDirectory: true,
+        files: MLPACKAGE_FILES,
+        runtimes: ["python"]
+      },
+      openvino: {
+        url: (0, import_types2.hfModelUrl)(HF_REPO2, "segmentation/yolo11-seg/openvino/camstack-yolo11n-seg.xml"),
+        sizeMB: 6,
+        runtimes: ["python"]
+      }
     }
-    function computeRawMask(coeffs, protos, numMaskCoeffs, maskH, maskW) {
-      const maskSize = maskH * maskW;
-      const rawMask = new Float32Array(maskSize);
-      for (let px = 0; px < maskSize; px++) {
-        let val = 0;
-        for (let k = 0; k < numMaskCoeffs; k++) {
-          val += (coeffs[k] ?? 0) * (protos[k * maskSize + px] ?? 0);
-        }
-        rawMask[px] = sigmoid(val);
+  },
+  {
+    id: "yolo11s-seg",
+    name: "YOLO11 Small Segmentation",
+    description: "YOLO11 Small \u2014 balanced speed and accuracy for instance segmentation",
+    inputSize: { width: 640, height: 640 },
+    labels: import_types2.COCO_80_LABELS,
+    formats: {
+      onnx: {
+        url: (0, import_types2.hfModelUrl)(HF_REPO2, "segmentation/yolo11-seg/onnx/camstack-yolo11s-seg.onnx"),
+        sizeMB: 39
+      },
+      coreml: {
+        url: (0, import_types2.hfModelUrl)(HF_REPO2, "segmentation/yolo11-seg/coreml/camstack-yolo11s-seg.mlpackage"),
+        sizeMB: 20,
+        isDirectory: true,
+        files: MLPACKAGE_FILES,
+        runtimes: ["python"]
+      },
+      openvino: {
+        url: (0, import_types2.hfModelUrl)(HF_REPO2, "segmentation/yolo11-seg/openvino/camstack-yolo11s-seg.xml"),
+        sizeMB: 20,
+        runtimes: ["python"]
       }
-      return rawMask;
     }
-    function cropAndThresholdMask(rawMask, maskH, maskW, bbox, maskThreshold, maskScale) {
-      const cropX1 = Math.max(0, Math.floor(bbox.x * maskScale));
-      const cropY1 = Math.max(0, Math.floor(bbox.y * maskScale));
-      const cropX2 = Math.min(maskW, Math.ceil((bbox.x + bbox.w) * maskScale));
-      const cropY2 = Math.min(maskH, Math.ceil((bbox.y + bbox.h) * maskScale));
-      const cropW = Math.max(1, cropX2 - cropX1);
-      const cropH = Math.max(1, cropY2 - cropY1);
-      const data = new Uint8Array(cropW * cropH);
-      for (let row = 0; row < cropH; row++) {
-        const srcRow = cropY1 + row;
-        for (let col = 0; col < cropW; col++) {
-          const srcCol = cropX1 + col;
-          const srcIdx = srcRow * maskW + srcCol;
-          data[row * cropW + col] = (rawMask[srcIdx] ?? 0) > maskThreshold ? 255 : 0;
-        }
+  },
+  {
+    id: "yolo11m-seg",
+    name: "YOLO11 Medium Segmentation",
+    description: "YOLO11 Medium \u2014 higher accuracy instance segmentation",
+    inputSize: { width: 640, height: 640 },
+    labels: import_types2.COCO_80_LABELS,
+    formats: {
+      onnx: {
+        url: (0, import_types2.hfModelUrl)(HF_REPO2, "segmentation/yolo11-seg/onnx/camstack-yolo11m-seg.onnx"),
+        sizeMB: 86
+      },
+      coreml: {
+        url: (0, import_types2.hfModelUrl)(HF_REPO2, "segmentation/yolo11-seg/coreml/camstack-yolo11m-seg.mlpackage"),
+        sizeMB: 43,
+        isDirectory: true,
+        files: MLPACKAGE_FILES,
+        runtimes: ["python"]
+      },
+      openvino: {
+        url: (0, import_types2.hfModelUrl)(HF_REPO2, "segmentation/yolo11-seg/openvino/camstack-yolo11m-seg.xml"),
+        sizeMB: 43,
+        runtimes: ["python"]
       }
-      return { data, width: cropW, height: cropH };
     }
-    function yoloSegPostprocess2(segOutput, options) {
-      const { detectionOutput, protoOutput, numClasses, numBoxes, numMaskCoeffs, maskHeight, maskWidth } = segOutput;
-      const { confidence, iouThreshold, labels, scale, padX, padY, originalWidth, originalHeight, maskThreshold = 0.5 } = options;
-      const yoloInputSize = 640;
-      const maskScale = maskHeight / yoloInputSize;
-      const candidates = [];
-      for (let i = 0; i < numBoxes; i++) {
-        const cx = detectionOutput[0 * numBoxes + i] ?? 0;
-        const cy = detectionOutput[1 * numBoxes + i] ?? 0;
-        const w = detectionOutput[2 * numBoxes + i] ?? 0;
-        const h = detectionOutput[3 * numBoxes + i] ?? 0;
-        let bestScore = -Infinity;
-        let bestClass = 0;
-        for (let j = 0; j < numClasses; j++) {
-          const score = detectionOutput[(4 + j) * numBoxes + i] ?? 0;
-          if (score > bestScore) {
-            bestScore = score;
-            bestClass = j;
-          }
-        }
-        if (bestScore < confidence)
-          continue;
-        const bbox = {
-          x: cx - w / 2,
-          y: cy - h / 2,
-          w,
-          h
-        };
-        const coeffs = new Float32Array(numMaskCoeffs);
-        for (let k = 0; k < numMaskCoeffs; k++) {
-          coeffs[k] = detectionOutput[(4 + numClasses + k) * numBoxes + i] ?? 0;
-        }
-        candidates.push({ bbox, score: bestScore, classIdx: bestClass, coeffs });
+  },
+  // ── YOLOv8-seg — CoreML supported ─────────────────────────────
+  {
+    id: "yolov8n-seg",
+    name: "YOLOv8 Nano Segmentation",
+    description: "YOLOv8 Nano \u2014 fastest, smallest YOLOv8 instance segmentation model",
+    inputSize: { width: 640, height: 640 },
+    labels: import_types2.COCO_80_LABELS,
+    formats: {
+      onnx: {
+        url: (0, import_types2.hfModelUrl)(HF_REPO2, "segmentation/yolov8-seg/onnx/camstack-yolov8n-seg.onnx"),
+        sizeMB: 13
+      },
+      coreml: {
+        url: (0, import_types2.hfModelUrl)(HF_REPO2, "segmentation/yolov8-seg/coreml/camstack-yolov8n-seg.mlpackage"),
+        sizeMB: 7,
+        isDirectory: true,
+        files: MLPACKAGE_FILES,
+        runtimes: ["python"]
+      },
+      openvino: {
+        url: (0, import_types2.hfModelUrl)(HF_REPO2, "segmentation/yolov8-seg/openvino/camstack-yolov8n-seg.xml"),
+        sizeMB: 7,
+        runtimes: ["python"]
+      }
+    }
+  },
+  {
+    id: "yolov8s-seg",
+    name: "YOLOv8 Small Segmentation",
+    description: "YOLOv8 Small \u2014 balanced speed and accuracy for instance segmentation",
+    inputSize: { width: 640, height: 640 },
+    labels: import_types2.COCO_80_LABELS,
+    formats: {
+      onnx: {
+        url: (0, import_types2.hfModelUrl)(HF_REPO2, "segmentation/yolov8-seg/onnx/camstack-yolov8s-seg.onnx"),
+        sizeMB: 45
+      },
+      coreml: {
+        url: (0, import_types2.hfModelUrl)(HF_REPO2, "segmentation/yolov8-seg/coreml/camstack-yolov8s-seg.mlpackage"),
+        sizeMB: 23,
+        isDirectory: true,
+        files: MLPACKAGE_FILES,
+        runtimes: ["python"]
+      },
+      openvino: {
+        url: (0, import_types2.hfModelUrl)(HF_REPO2, "segmentation/yolov8-seg/openvino/camstack-yolov8s-seg.xml"),
+        sizeMB: 23,
+        runtimes: ["python"]
+      }
+    }
+  },
+  {
+    id: "yolov8m-seg",
+    name: "YOLOv8 Medium Segmentation",
+    description: "YOLOv8 Medium \u2014 higher accuracy instance segmentation",
+    inputSize: { width: 640, height: 640 },
+    labels: import_types2.COCO_80_LABELS,
+    formats: {
+      onnx: {
+        url: (0, import_types2.hfModelUrl)(HF_REPO2, "segmentation/yolov8-seg/onnx/camstack-yolov8m-seg.onnx"),
+        sizeMB: 104
+      },
+      coreml: {
+        url: (0, import_types2.hfModelUrl)(HF_REPO2, "segmentation/yolov8-seg/coreml/camstack-yolov8m-seg.mlpackage"),
+        sizeMB: 52,
+        isDirectory: true,
+        files: MLPACKAGE_FILES,
+        runtimes: ["python"]
+      },
+      openvino: {
+        url: (0, import_types2.hfModelUrl)(HF_REPO2, "segmentation/yolov8-seg/openvino/camstack-yolov8m-seg.xml"),
+        sizeMB: 53,
+        runtimes: ["python"]
       }
-      if (candidates.length === 0)
-        return [];
-      const keptIndices = (0, yolo_js_1.nms)(candidates, iouThreshold);
-      return keptIndices.map((idx) => {
-        const { bbox, score, classIdx, coeffs } = candidates[idx];
-        const label = labels[classIdx] ?? String(classIdx);
-        const x = Math.max(0, Math.min(originalWidth, (bbox.x - padX) / scale));
-        const y = Math.max(0, Math.min(originalHeight, (bbox.y - padY) / scale));
-        const x2 = Math.max(0, Math.min(originalWidth, (bbox.x + bbox.w - padX) / scale));
-        const y2 = Math.max(0, Math.min(originalHeight, (bbox.y + bbox.h - padY) / scale));
-        const finalBbox = { x, y, w: x2 - x, h: y2 - y };
-        const rawMask = computeRawMask(coeffs, protoOutput, numMaskCoeffs, maskHeight, maskWidth);
-        const { data: maskData, width: mW, height: mH } = cropAndThresholdMask(rawMask, maskHeight, maskWidth, bbox, maskThreshold, maskScale);
-        return {
-          class: label,
-          originalClass: label,
-          score,
-          bbox: finalBbox,
-          mask: maskData,
-          maskWidth: mW,
-          maskHeight: mH
-        };
-      });
     }
   }
-});
+];
+
+// src/addons/object-detection/index.ts
+var import_types3 = require("@camstack/types");
 
-// src/shared/node-engine.js
-var require_node_engine = __commonJS({
-  "src/shared/node-engine.js"(exports2) {
-    "use strict";
-    var __createBinding = exports2 && exports2.__createBinding || (Object.create ? (function(o, m, k, k2) {
-      if (k2 === void 0) k2 = k;
-      var desc = Object.getOwnPropertyDescriptor(m, k);
-      if (!desc || ("get" in desc ? !m.__esModule : desc.writable || desc.configurable)) {
-        desc = { enumerable: true, get: function() {
-          return m[k];
-        } };
+// src/shared/image-utils.ts
+var import_sharp = __toESM(require("sharp"));
+async function letterbox(jpeg, targetSize) {
+  const meta = await (0, import_sharp.default)(jpeg).metadata();
+  const originalWidth = meta.width ?? 0;
+  const originalHeight = meta.height ?? 0;
+  const scale = Math.min(targetSize / originalWidth, targetSize / originalHeight);
+  const scaledWidth = Math.round(originalWidth * scale);
+  const scaledHeight = Math.round(originalHeight * scale);
+  const padX = Math.floor((targetSize - scaledWidth) / 2);
+  const padY = Math.floor((targetSize - scaledHeight) / 2);
+  const { data } = await (0, import_sharp.default)(jpeg).resize(scaledWidth, scaledHeight).extend({
+    top: padY,
+    bottom: targetSize - scaledHeight - padY,
+    left: padX,
+    right: targetSize - scaledWidth - padX,
+    background: { r: 114, g: 114, b: 114 }
+  }).removeAlpha().raw().toBuffer({ resolveWithObject: true });
+  const numPixels = targetSize * targetSize;
+  const float32 = new Float32Array(3 * numPixels);
+  for (let i = 0; i < numPixels; i++) {
+    const srcBase = i * 3;
+    float32[0 * numPixels + i] = data[srcBase] / 255;
+    float32[1 * numPixels + i] = data[srcBase + 1] / 255;
+    float32[2 * numPixels + i] = data[srcBase + 2] / 255;
+  }
+  return { data: float32, scale, padX, padY, originalWidth, originalHeight };
+}
+
+// src/shared/postprocess/yolo.ts
+function iou(a, b) {
+  const ax1 = a.x;
+  const ay1 = a.y;
+  const ax2 = a.x + a.w;
+  const ay2 = a.y + a.h;
+  const bx1 = b.x;
+  const by1 = b.y;
+  const bx2 = b.x + b.w;
+  const by2 = b.y + b.h;
+  const interX1 = Math.max(ax1, bx1);
+  const interY1 = Math.max(ay1, by1);
+  const interX2 = Math.min(ax2, bx2);
+  const interY2 = Math.min(ay2, by2);
+  const interW = Math.max(0, interX2 - interX1);
+  const interH = Math.max(0, interY2 - interY1);
+  const interArea = interW * interH;
+  if (interArea === 0) return 0;
+  const areaA = a.w * a.h;
+  const areaB = b.w * b.h;
+  const unionArea = areaA + areaB - interArea;
+  return unionArea === 0 ? 0 : interArea / unionArea;
+}
+function nms(boxes, iouThreshold) {
+  const indices = boxes.map((_, i) => i).sort((a, b) => boxes[b].score - boxes[a].score);
+  const kept = [];
+  const suppressed = /* @__PURE__ */ new Set();
+  for (const idx of indices) {
+    if (suppressed.has(idx)) continue;
+    kept.push(idx);
+    for (const other of indices) {
+      if (other === idx || suppressed.has(other)) continue;
+      if (iou(boxes[idx].bbox, boxes[other].bbox) > iouThreshold) {
+        suppressed.add(other);
       }
-      Object.defineProperty(o, k2, desc);
-    }) : (function(o, m, k, k2) {
-      if (k2 === void 0) k2 = k;
-      o[k2] = m[k];
-    }));
-    var __setModuleDefault = exports2 && exports2.__setModuleDefault || (Object.create ? (function(o, v) {
-      Object.defineProperty(o, "default", { enumerable: true, value: v });
-    }) : function(o, v) {
-      o["default"] = v;
-    });
-    var __importStar = exports2 && exports2.__importStar || /* @__PURE__ */ (function() {
-      var ownKeys = function(o) {
-        ownKeys = Object.getOwnPropertyNames || function(o2) {
-          var ar = [];
-          for (var k in o2) if (Object.prototype.hasOwnProperty.call(o2, k)) ar[ar.length] = k;
-          return ar;
-        };
-        return ownKeys(o);
-      };
-      return function(mod) {
-        if (mod && mod.__esModule) return mod;
-        var result = {};
-        if (mod != null) {
-          for (var k = ownKeys(mod), i = 0; i < k.length; i++) if (k[i] !== "default") __createBinding(result, mod, k[i]);
-        }
-        __setModuleDefault(result, mod);
-        return result;
-      };
-    })();
-    Object.defineProperty(exports2, "__esModule", { value: true });
-    exports2.NodeInferenceEngine = void 0;
-    var path = __importStar(require("path"));
-    var BACKEND_TO_PROVIDER = {
-      cpu: "cpu",
-      coreml: "coreml",
-      cuda: "cuda",
-      tensorrt: "tensorrt",
-      dml: "dml"
+    }
+  }
+  return kept;
+}
+function yoloPostprocess(output, numClasses, numBoxes, options) {
+  const { confidence, iouThreshold, labels, scale, padX, padY, originalWidth, originalHeight } = options;
+  const candidates = [];
+  for (let i = 0; i < numBoxes; i++) {
+    const cx = output[0 * numBoxes + i];
+    const cy = output[1 * numBoxes + i];
+    const w = output[2 * numBoxes + i];
+    const h = output[3 * numBoxes + i];
+    let bestScore = -Infinity;
+    let bestClass = 0;
+    for (let j = 0; j < numClasses; j++) {
+      const score = output[(4 + j) * numBoxes + i];
+      if (score > bestScore) {
+        bestScore = score;
+        bestClass = j;
+      }
+    }
+    if (bestScore < confidence) continue;
+    const bbox = {
+      x: cx - w / 2,
+      y: cy - h / 2,
+      w,
+      h
     };
-    var BACKEND_TO_DEVICE = {
-      cpu: "cpu",
-      coreml: "gpu-mps",
-      cuda: "gpu-cuda",
-      tensorrt: "tensorrt"
+    candidates.push({ bbox, score: bestScore, classIdx: bestClass });
+  }
+  if (candidates.length === 0) return [];
+  const keptIndices = nms(candidates, iouThreshold);
+  return keptIndices.map((idx) => {
+    const { bbox, score, classIdx } = candidates[idx];
+    const label = labels[classIdx] ?? String(classIdx);
+    const x = Math.max(0, Math.min(originalWidth, (bbox.x - padX) / scale));
+    const y = Math.max(0, Math.min(originalHeight, (bbox.y - padY) / scale));
+    const x2 = Math.max(0, Math.min(originalWidth, (bbox.x + bbox.w - padX) / scale));
+    const y2 = Math.max(0, Math.min(originalHeight, (bbox.y + bbox.h - padY) / scale));
+    const finalBbox = { x, y, w: x2 - x, h: y2 - y };
+    return {
+      class: label,
+      originalClass: label,
+      score,
+      bbox: finalBbox
     };
-    var NodeInferenceEngine = class {
-      modelPath;
-      backend;
-      runtime = "onnx";
-      device;
-      session = null;
-      constructor(modelPath, backend) {
-        this.modelPath = modelPath;
-        this.backend = backend;
-        this.device = BACKEND_TO_DEVICE[backend] ?? "cpu";
-      }
-      async initialize() {
-        const ort = await Promise.resolve().then(() => __importStar(require("onnxruntime-node")));
-        const provider = BACKEND_TO_PROVIDER[this.backend] ?? "cpu";
-        const absModelPath = path.isAbsolute(this.modelPath) ? this.modelPath : path.resolve(process.cwd(), this.modelPath);
-        const sessionOptions = {
-          executionProviders: [provider]
-        };
-        this.session = await ort.InferenceSession.create(absModelPath, sessionOptions);
-      }
-      async run(input, inputShape) {
-        if (!this.session) {
-          throw new Error("NodeInferenceEngine: not initialized \u2014 call initialize() first");
-        }
-        const ort = await Promise.resolve().then(() => __importStar(require("onnxruntime-node")));
-        const sess = this.session;
-        const inputName = sess.inputNames[0];
-        const tensor = new ort.Tensor("float32", input, [...inputShape]);
-        const feeds = { [inputName]: tensor };
-        const results = await sess.run(feeds);
-        const outputName = sess.outputNames[0];
-        const outputTensor = results[outputName];
-        return outputTensor.data;
-      }
-      async runMultiOutput(input, inputShape) {
-        if (!this.session) {
-          throw new Error("NodeInferenceEngine: not initialized \u2014 call initialize() first");
-        }
-        const ort = await Promise.resolve().then(() => __importStar(require("onnxruntime-node")));
-        const sess = this.session;
-        const inputName = sess.inputNames[0];
-        const tensor = new ort.Tensor("float32", input, [...inputShape]);
-        const feeds = { [inputName]: tensor };
-        const results = await sess.run(feeds);
-        const out = {};
-        for (const name of sess.outputNames) {
-          out[name] = results[name].data;
-        }
-        return out;
-      }
-      async dispose() {
-        this.session = null;
+  });
+}
+
+// src/shared/postprocess/yolo-seg.ts
+function sigmoid(x) {
+  return 1 / (1 + Math.exp(-x));
+}
+function computeRawMask(coeffs, protos, numMaskCoeffs, maskH, maskW) {
+  const maskSize = maskH * maskW;
+  const rawMask = new Float32Array(maskSize);
+  for (let px = 0; px < maskSize; px++) {
+    let val = 0;
+    for (let k = 0; k < numMaskCoeffs; k++) {
+      val += (coeffs[k] ?? 0) * (protos[k * maskSize + px] ?? 0);
+    }
+    rawMask[px] = sigmoid(val);
+  }
+  return rawMask;
+}
+function cropAndThresholdMask(rawMask, maskH, maskW, bbox, maskThreshold, maskScale) {
+  const cropX1 = Math.max(0, Math.floor(bbox.x * maskScale));
+  const cropY1 = Math.max(0, Math.floor(bbox.y * maskScale));
+  const cropX2 = Math.min(maskW, Math.ceil((bbox.x + bbox.w) * maskScale));
+  const cropY2 = Math.min(maskH, Math.ceil((bbox.y + bbox.h) * maskScale));
+  const cropW = Math.max(1, cropX2 - cropX1);
+  const cropH = Math.max(1, cropY2 - cropY1);
+  const data = new Uint8Array(cropW * cropH);
+  for (let row = 0; row < cropH; row++) {
+    const srcRow = cropY1 + row;
+    for (let col = 0; col < cropW; col++) {
+      const srcCol = cropX1 + col;
+      const srcIdx = srcRow * maskW + srcCol;
+      data[row * cropW + col] = (rawMask[srcIdx] ?? 0) > maskThreshold ? 255 : 0;
+    }
+  }
+  return { data, width: cropW, height: cropH };
+}
+function yoloSegPostprocess(segOutput, options) {
+  const {
+    detectionOutput,
+    protoOutput,
+    numClasses,
+    numBoxes,
+    numMaskCoeffs,
+    maskHeight,
+    maskWidth
+  } = segOutput;
+  const {
+    confidence,
+    iouThreshold,
+    labels,
+    scale,
+    padX,
+    padY,
+    originalWidth,
+    originalHeight,
+    maskThreshold = 0.5
+  } = options;
+  const yoloInputSize = 640;
+  const maskScale = maskHeight / yoloInputSize;
+  const candidates = [];
+  for (let i = 0; i < numBoxes; i++) {
+    const cx = detectionOutput[0 * numBoxes + i] ?? 0;
+    const cy = detectionOutput[1 * numBoxes + i] ?? 0;
+    const w = detectionOutput[2 * numBoxes + i] ?? 0;
+    const h = detectionOutput[3 * numBoxes + i] ?? 0;
+    let bestScore = -Infinity;
+    let bestClass = 0;
+    for (let j = 0; j < numClasses; j++) {
+      const score = detectionOutput[(4 + j) * numBoxes + i] ?? 0;
+      if (score > bestScore) {
+        bestScore = score;
+        bestClass = j;
       }
+    }
+    if (bestScore < confidence) continue;
+    const bbox = {
+      x: cx - w / 2,
+      y: cy - h / 2,
+      w,
+      h
     };
-    exports2.NodeInferenceEngine = NodeInferenceEngine;
+    const coeffs = new Float32Array(numMaskCoeffs);
+    for (let k = 0; k < numMaskCoeffs; k++) {
+      coeffs[k] = detectionOutput[(4 + numClasses + k) * numBoxes + i] ?? 0;
+    }
+    candidates.push({ bbox, score: bestScore, classIdx: bestClass, coeffs });
   }
-});
+  if (candidates.length === 0) return [];
+  const keptIndices = nms(candidates, iouThreshold);
+  return keptIndices.map((idx) => {
+    const { bbox, score, classIdx, coeffs } = candidates[idx];
+    const label = labels[classIdx] ?? String(classIdx);
+    const x = Math.max(0, Math.min(originalWidth, (bbox.x - padX) / scale));
+    const y = Math.max(0, Math.min(originalHeight, (bbox.y - padY) / scale));
+    const x2 = Math.max(0, Math.min(originalWidth, (bbox.x + bbox.w - padX) / scale));
+    const y2 = Math.max(0, Math.min(originalHeight, (bbox.y + bbox.h - padY) / scale));
+    const finalBbox = { x, y, w: x2 - x, h: y2 - y };
+    const rawMask = computeRawMask(coeffs, protoOutput, numMaskCoeffs, maskHeight, maskWidth);
+    const { data: maskData, width: mW, height: mH } = cropAndThresholdMask(
+      rawMask,
+      maskHeight,
+      maskWidth,
+      bbox,
+      maskThreshold,
+      maskScale
+    );
+    return {
+      class: label,
+      originalClass: label,
+      score,
+      bbox: finalBbox,
+      mask: maskData,
+      maskWidth: mW,
+      maskHeight: mH
+    };
+  });
+}
 
-// src/shared/python-engine.js
-var require_python_engine = __commonJS({
-  "src/shared/python-engine.js"(exports2) {
-    "use strict";
-    Object.defineProperty(exports2, "__esModule", { value: true });
-    exports2.PythonInferenceEngine = void 0;
-    exports2.resolvePythonBinary = resolvePythonBinary;
-    var node_child_process_1 = require("child_process");
-    var PythonInferenceEngine = class {
-      pythonPath;
-      scriptPath;
-      modelPath;
-      extraArgs;
-      runtime;
-      device;
-      process = null;
-      receiveBuffer = Buffer.alloc(0);
-      pendingResolve = null;
-      pendingReject = null;
-      constructor(pythonPath, scriptPath, runtime, modelPath, extraArgs = []) {
-        this.pythonPath = pythonPath;
-        this.scriptPath = scriptPath;
-        this.modelPath = modelPath;
-        this.extraArgs = extraArgs;
-        this.runtime = runtime;
-        const runtimeDeviceMap = {
-          onnx: "cpu",
-          coreml: "gpu-mps",
-          pytorch: "cpu",
-          openvino: "cpu",
-          tflite: "cpu"
-        };
-        this.device = runtimeDeviceMap[runtime];
-      }
-      async initialize() {
-        const args = [this.scriptPath, this.modelPath, ...this.extraArgs];
-        this.process = (0, node_child_process_1.spawn)(this.pythonPath, args, {
-          stdio: ["pipe", "pipe", "pipe"]
-        });
-        if (!this.process.stdout || !this.process.stdin) {
-          throw new Error("PythonInferenceEngine: failed to create process pipes");
-        }
-        this.process.stderr?.on("data", (chunk) => {
-          process.stderr.write(`[python-engine] ${chunk.toString()}`);
-        });
-        this.process.on("error", (err) => {
-          this.pendingReject?.(err);
-          this.pendingReject = null;
-          this.pendingResolve = null;
-        });
-        this.process.on("exit", (code) => {
-          if (code !== 0) {
-            const err = new Error(`PythonInferenceEngine: process exited with code ${code}`);
-            this.pendingReject?.(err);
-            this.pendingReject = null;
-            this.pendingResolve = null;
-          }
-        });
-        this.process.stdout.on("data", (chunk) => {
-          this.receiveBuffer = Buffer.concat([this.receiveBuffer, chunk]);
-          this._tryReceive();
-        });
-        await new Promise((resolve, reject) => {
-          const timeout = setTimeout(() => resolve(), 2e3);
-          this.process?.on("error", (err) => {
-            clearTimeout(timeout);
-            reject(err);
-          });
-          this.process?.on("exit", (code) => {
-            clearTimeout(timeout);
-            if (code !== 0) {
-              reject(new Error(`PythonInferenceEngine: process exited early with code ${code}`));
-            }
-          });
-        });
-      }
-      _tryReceive() {
-        if (this.receiveBuffer.length < 4)
-          return;
-        const length = this.receiveBuffer.readUInt32LE(0);
-        if (this.receiveBuffer.length < 4 + length)
-          return;
-        const jsonBytes = this.receiveBuffer.subarray(4, 4 + length);
-        this.receiveBuffer = this.receiveBuffer.subarray(4 + length);
-        const resolve = this.pendingResolve;
-        const reject = this.pendingReject;
-        this.pendingResolve = null;
-        this.pendingReject = null;
-        if (!resolve)
-          return;
-        try {
-          const parsed = JSON.parse(jsonBytes.toString("utf8"));
-          resolve(parsed);
-        } catch (err) {
-          reject?.(err instanceof Error ? err : new Error(String(err)));
-        }
-      }
-      /** Send JPEG buffer, receive JSON detection results */
-      async runJpeg(jpeg) {
-        if (!this.process?.stdin) {
-          throw new Error("PythonInferenceEngine: process not initialized");
-        }
-        return new Promise((resolve, reject) => {
-          this.pendingResolve = resolve;
-          this.pendingReject = reject;
-          const lengthBuf = Buffer.allocUnsafe(4);
-          lengthBuf.writeUInt32LE(jpeg.length, 0);
-          this.process.stdin.write(Buffer.concat([lengthBuf, jpeg]));
-        });
-      }
-      /** IInferenceEngine.run — wraps runJpeg for compatibility */
-      async run(_input, _inputShape) {
-        throw new Error("PythonInferenceEngine: use runJpeg() directly \u2014 this engine operates on JPEG input");
-      }
-      /** IInferenceEngine.runMultiOutput — not supported by Python engine (operates on JPEG input) */
-      async runMultiOutput(_input, _inputShape) {
-        throw new Error("PythonInferenceEngine: runMultiOutput() is not supported \u2014 this engine operates on JPEG input");
-      }
-      async dispose() {
-        if (this.process) {
-          this.process.stdin?.end();
-          this.process.kill("SIGTERM");
-          this.process = null;
-        }
-      }
+// src/shared/engine-resolver.ts
+var fs = __toESM(require("fs"));
+var path2 = __toESM(require("path"));
+
+// src/shared/node-engine.ts
+var path = __toESM(require("path"));
+var BACKEND_TO_PROVIDER = {
+  cpu: "cpu",
+  coreml: "coreml",
+  cuda: "cuda",
+  tensorrt: "tensorrt",
+  dml: "dml"
+};
+var BACKEND_TO_DEVICE = {
+  cpu: "cpu",
+  coreml: "gpu-mps",
+  cuda: "gpu-cuda",
+  tensorrt: "tensorrt"
+};
+var NodeInferenceEngine = class {
+  constructor(modelPath, backend) {
+    this.modelPath = modelPath;
+    this.backend = backend;
+    this.device = BACKEND_TO_DEVICE[backend] ?? "cpu";
+  }
+  runtime = "onnx";
+  device;
+  session = null;
+  async initialize() {
+    const ort = await import("onnxruntime-node");
+    const provider = BACKEND_TO_PROVIDER[this.backend] ?? "cpu";
+    const absModelPath = path.isAbsolute(this.modelPath) ? this.modelPath : path.resolve(process.cwd(), this.modelPath);
+    const sessionOptions = {
+      executionProviders: [provider]
     };
-    exports2.PythonInferenceEngine = PythonInferenceEngine;
-    async function resolvePythonBinary(configPath, deps) {
-      if (configPath)
-        return configPath;
-      return deps.ensurePython();
+    this.session = await ort.InferenceSession.create(absModelPath, sessionOptions);
+  }
+  async run(input, inputShape) {
+    if (!this.session) {
+      throw new Error("NodeInferenceEngine: not initialized \u2014 call initialize() first");
     }
+    const ort = await import("onnxruntime-node");
+    const sess = this.session;
+    const inputName = sess.inputNames[0];
+    const tensor = new ort.Tensor("float32", input, [...inputShape]);
+    const feeds = { [inputName]: tensor };
+    const results = await sess.run(feeds);
+    const outputName = sess.outputNames[0];
+    const outputTensor = results[outputName];
+    return outputTensor.data;
   }
-});
+  async runMultiOutput(input, inputShape) {
+    if (!this.session) {
+      throw new Error("NodeInferenceEngine: not initialized \u2014 call initialize() first");
+    }
+    const ort = await import("onnxruntime-node");
+    const sess = this.session;
+    const inputName = sess.inputNames[0];
+    const tensor = new ort.Tensor("float32", input, [...inputShape]);
+    const feeds = { [inputName]: tensor };
+    const results = await sess.run(feeds);
+    const out = {};
+    for (const name of sess.outputNames) {
+      out[name] = results[name].data;
+    }
+    return out;
+  }
+  async dispose() {
+    this.session = null;
+  }
+};
 
-// src/shared/engine-resolver.js
-var require_engine_resolver = __commonJS({
-  "src/shared/engine-resolver.js"(exports2) {
-    "use strict";
-    var __createBinding = exports2 && exports2.__createBinding || (Object.create ? (function(o, m, k, k2) {
-      if (k2 === void 0) k2 = k;
-      var desc = Object.getOwnPropertyDescriptor(m, k);
-      if (!desc || ("get" in desc ? !m.__esModule : desc.writable || desc.configurable)) {
-        desc = { enumerable: true, get: function() {
-          return m[k];
-        } };
+// src/shared/python-engine.ts
+var import_node_child_process = require("child_process");
+var PythonInferenceEngine = class {
+  constructor(pythonPath, scriptPath, runtime, modelPath, extraArgs = []) {
+    this.pythonPath = pythonPath;
+    this.scriptPath = scriptPath;
+    this.modelPath = modelPath;
+    this.extraArgs = extraArgs;
+    this.runtime = runtime;
+    const runtimeDeviceMap = {
+      onnx: "cpu",
+      coreml: "gpu-mps",
+      pytorch: "cpu",
+      openvino: "cpu",
+      tflite: "cpu"
+    };
+    this.device = runtimeDeviceMap[runtime];
+  }
+  runtime;
+  device;
+  process = null;
+  receiveBuffer = Buffer.alloc(0);
+  pendingResolve = null;
+  pendingReject = null;
+  async initialize() {
+    const args = [this.scriptPath, this.modelPath, ...this.extraArgs];
+    this.process = (0, import_node_child_process.spawn)(this.pythonPath, args, {
+      stdio: ["pipe", "pipe", "pipe"]
+    });
+    if (!this.process.stdout || !this.process.stdin) {
+      throw new Error("PythonInferenceEngine: failed to create process pipes");
+    }
+    this.process.stderr?.on("data", (chunk) => {
+      process.stderr.write(`[python-engine] ${chunk.toString()}`);
+    });
+    this.process.on("error", (err) => {
+      this.pendingReject?.(err);
+      this.pendingReject = null;
+      this.pendingResolve = null;
+    });
+    this.process.on("exit", (code) => {
+      if (code !== 0) {
+        const err = new Error(`PythonInferenceEngine: process exited with code ${code}`);
+        this.pendingReject?.(err);
+        this.pendingReject = null;
+        this.pendingResolve = null;
       }
-      Object.defineProperty(o, k2, desc);
-    }) : (function(o, m, k, k2) {
-      if (k2 === void 0) k2 = k;
-      o[k2] = m[k];
-    }));
-    var __setModuleDefault = exports2 && exports2.__setModuleDefault || (Object.create ? (function(o, v) {
-      Object.defineProperty(o, "default", { enumerable: true, value: v });
-    }) : function(o, v) {
-      o["default"] = v;
     });
-    var __importStar = exports2 && exports2.__importStar || /* @__PURE__ */ (function() {
-      var ownKeys = function(o) {
-        ownKeys = Object.getOwnPropertyNames || function(o2) {
-          var ar = [];
-          for (var k in o2) if (Object.prototype.hasOwnProperty.call(o2, k)) ar[ar.length] = k;
-          return ar;
-        };
-        return ownKeys(o);
-      };
-      return function(mod) {
-        if (mod && mod.__esModule) return mod;
-        var result = {};
-        if (mod != null) {
-          for (var k = ownKeys(mod), i = 0; i < k.length; i++) if (k[i] !== "default") __createBinding(result, mod, k[i]);
+    this.process.stdout.on("data", (chunk) => {
+      this.receiveBuffer = Buffer.concat([this.receiveBuffer, chunk]);
+      this._tryReceive();
+    });
+    await new Promise((resolve2, reject) => {
+      const timeout = setTimeout(() => resolve2(), 2e3);
+      this.process?.on("error", (err) => {
+        clearTimeout(timeout);
+        reject(err);
+      });
+      this.process?.on("exit", (code) => {
+        clearTimeout(timeout);
+        if (code !== 0) {
+          reject(new Error(`PythonInferenceEngine: process exited early with code ${code}`));
         }
-        __setModuleDefault(result, mod);
-        return result;
-      };
-    })();
-    Object.defineProperty(exports2, "__esModule", { value: true });
-    exports2.resolveEngine = resolveEngine2;
-    exports2.probeOnnxBackends = probeOnnxBackends;
-    var fs = __importStar(require("fs"));
-    var path = __importStar(require("path"));
-    var node_engine_js_1 = require_node_engine();
-    var python_engine_js_1 = require_python_engine();
-    var AUTO_BACKEND_PRIORITY = ["coreml", "cuda", "tensorrt", "cpu"];
-    var BACKEND_TO_FORMAT = {
-      cpu: "onnx",
-      coreml: "onnx",
-      cuda: "onnx",
-      tensorrt: "onnx"
-    };
-    var RUNTIME_TO_FORMAT = {
-      onnx: "onnx",
-      coreml: "coreml",
-      openvino: "openvino",
-      tflite: "tflite",
-      pytorch: "pt"
-    };
-    function modelFilePath(modelsDir, modelEntry, format) {
-      const formatEntry = modelEntry.formats[format];
-      if (!formatEntry) {
-        throw new Error(`Model ${modelEntry.id} has no ${format} format`);
-      }
-      const urlParts = formatEntry.url.split("/");
-      const filename = urlParts[urlParts.length - 1] ?? `${modelEntry.id}.${format}`;
-      return path.join(modelsDir, filename);
+      });
+    });
+  }
+  _tryReceive() {
+    if (this.receiveBuffer.length < 4) return;
+    const length = this.receiveBuffer.readUInt32LE(0);
+    if (this.receiveBuffer.length < 4 + length) return;
+    const jsonBytes = this.receiveBuffer.subarray(4, 4 + length);
+    this.receiveBuffer = this.receiveBuffer.subarray(4 + length);
+    const resolve2 = this.pendingResolve;
+    const reject = this.pendingReject;
+    this.pendingResolve = null;
+    this.pendingReject = null;
+    if (!resolve2) return;
+    try {
+      const parsed = JSON.parse(jsonBytes.toString("utf8"));
+      resolve2(parsed);
+    } catch (err) {
+      reject?.(err instanceof Error ? err : new Error(String(err)));
     }
-    function modelExists(filePath) {
-      try {
-        return fs.existsSync(filePath);
-      } catch {
-        return false;
-      }
+  }
+  /** Send JPEG buffer, receive JSON detection results */
+  async runJpeg(jpeg) {
+    if (!this.process?.stdin) {
+      throw new Error("PythonInferenceEngine: process not initialized");
     }
-    async function resolveEngine2(options) {
-      const { runtime, backend, modelEntry, modelsDir, models } = options;
-      let selectedFormat;
-      let selectedBackend;
-      if (runtime === "auto") {
-        const available = await probeOnnxBackends();
-        let chosen = null;
-        for (const b of AUTO_BACKEND_PRIORITY) {
-          if (!available.includes(b))
-            continue;
-          const fmt = BACKEND_TO_FORMAT[b];
-          if (!fmt)
-            continue;
-          if (!modelEntry.formats[fmt])
-            continue;
-          chosen = { backend: b, format: fmt };
-          break;
-        }
-        if (!chosen) {
-          throw new Error(`resolveEngine: no compatible backend found for model ${modelEntry.id}. Available backends: ${available.join(", ")}`);
-        }
-        selectedFormat = chosen.format;
-        selectedBackend = chosen.backend;
-      } else {
-        const fmt = RUNTIME_TO_FORMAT[runtime];
-        if (!fmt) {
-          throw new Error(`resolveEngine: unsupported runtime "${runtime}"`);
-        }
-        if (!modelEntry.formats[fmt]) {
-          throw new Error(`resolveEngine: model ${modelEntry.id} has no ${fmt} format for runtime ${runtime}`);
-        }
-        selectedFormat = fmt;
-        selectedBackend = runtime === "onnx" ? backend || "cpu" : runtime;
-      }
-      let modelPath;
-      if (models) {
-        modelPath = await models.ensure(modelEntry.id, selectedFormat);
-      } else {
-        modelPath = modelFilePath(modelsDir, modelEntry, selectedFormat);
-        if (!modelExists(modelPath)) {
-          throw new Error(`resolveEngine: model file not found at ${modelPath} and no model service provided`);
-        }
-      }
-      if (selectedFormat === "onnx") {
-        const engine = new node_engine_js_1.NodeInferenceEngine(modelPath, selectedBackend);
-        await engine.initialize();
-        return { engine, format: selectedFormat, modelPath };
-      }
-      const { pythonPath } = options;
-      const PYTHON_SCRIPT_MAP = {
-        coreml: "coreml_inference.py",
-        pytorch: "pytorch_inference.py",
-        openvino: "openvino_inference.py"
-      };
-      const effectiveRuntime = runtime === "auto" ? selectedBackend : runtime;
-      const scriptName = PYTHON_SCRIPT_MAP[effectiveRuntime];
-      if (scriptName && pythonPath) {
-        const candidates = [
-          path.join(__dirname, "../../python", scriptName),
-          path.join(__dirname, "../python", scriptName),
-          path.join(__dirname, "../../../python", scriptName)
-        ];
-        const scriptPath = candidates.find((p) => fs.existsSync(p));
-        if (!scriptPath) {
-          throw new Error(`resolveEngine: Python script "${scriptName}" not found. Searched:
-${candidates.join("\n")}`);
-        }
-        const inputSize = Math.max(modelEntry.inputSize.width, modelEntry.inputSize.height);
-        const engine = new python_engine_js_1.PythonInferenceEngine(pythonPath, scriptPath, effectiveRuntime, modelPath, [
-          `--input-size=${inputSize}`,
-          `--confidence=0.25`
-        ]);
-        await engine.initialize();
-        return { engine, format: selectedFormat, modelPath };
-      }
-      const fallbackPath = modelFilePath(modelsDir, modelEntry, "onnx");
-      if (modelEntry.formats["onnx"] && modelExists(fallbackPath)) {
-        const engine = new node_engine_js_1.NodeInferenceEngine(fallbackPath, "cpu");
-        await engine.initialize();
-        return { engine, format: "onnx", modelPath: fallbackPath };
-      }
-      throw new Error(`resolveEngine: format ${selectedFormat} is not yet supported by NodeInferenceEngine, no Python runtime is available, and no ONNX fallback exists`);
+    return new Promise((resolve2, reject) => {
+      this.pendingResolve = resolve2;
+      this.pendingReject = reject;
+      const lengthBuf = Buffer.allocUnsafe(4);
+      lengthBuf.writeUInt32LE(jpeg.length, 0);
+      this.process.stdin.write(Buffer.concat([lengthBuf, jpeg]));
+    });
+  }
+  /** IInferenceEngine.run — wraps runJpeg for compatibility */
+  async run(_input, _inputShape) {
+    throw new Error(
+      "PythonInferenceEngine: use runJpeg() directly \u2014 this engine operates on JPEG input"
+    );
+  }
+  /** IInferenceEngine.runMultiOutput — not supported by Python engine (operates on JPEG input) */
+  async runMultiOutput(_input, _inputShape) {
+    throw new Error(
+      "PythonInferenceEngine: runMultiOutput() is not supported \u2014 this engine operates on JPEG input"
+    );
+  }
+  async dispose() {
+    if (this.process) {
+      this.process.stdin?.end();
+      this.process.kill("SIGTERM");
+      this.process = null;
     }
-    async function probeOnnxBackends() {
-      const available = ["cpu"];
-      try {
-        const ort = await Promise.resolve().then(() => __importStar(require("onnxruntime-node")));
-        const providers = ort.env?.webgl?.disabled !== void 0 ? ort.InferenceSession?.getAvailableProviders?.() ?? [] : [];
-        for (const p of providers) {
-          const normalized = p.toLowerCase().replace("executionprovider", "");
-          if (normalized === "coreml")
-            available.push("coreml");
-          else if (normalized === "cuda")
-            available.push("cuda");
-          else if (normalized === "tensorrt")
-            available.push("tensorrt");
-        }
-      } catch {
-      }
-      if (process.platform === "darwin" && !available.includes("coreml")) {
-        available.push("coreml");
-      }
-      return [...new Set(available)];
+  }
+};
+
+// src/shared/engine-resolver.ts
+var AUTO_BACKEND_PRIORITY = ["coreml", "cuda", "tensorrt", "cpu"];
+var BACKEND_TO_FORMAT = {
+  cpu: "onnx",
+  coreml: "onnx",
+  cuda: "onnx",
+  tensorrt: "onnx"
+};
+var RUNTIME_TO_FORMAT = {
+  onnx: "onnx",
+  coreml: "coreml",
+  openvino: "openvino",
+  tflite: "tflite",
+  pytorch: "pt"
+};
+function modelFilePath(modelsDir, modelEntry, format) {
+  const formatEntry = modelEntry.formats[format];
+  if (!formatEntry) {
+    throw new Error(`Model ${modelEntry.id} has no ${format} format`);
+  }
+  const urlParts = formatEntry.url.split("/");
+  const filename = urlParts[urlParts.length - 1] ?? `${modelEntry.id}.${format}`;
+  return path2.join(modelsDir, filename);
+}
+function modelExists(filePath) {
+  try {
+    return fs.existsSync(filePath);
+  } catch {
+    return false;
+  }
+}
+async function resolveEngine(options) {
+  const { runtime, backend, modelEntry, modelsDir, models } = options;
+  let selectedFormat;
+  let selectedBackend;
+  if (runtime === "auto") {
+    const available = await probeOnnxBackends();
+    let chosen = null;
+    for (const b of AUTO_BACKEND_PRIORITY) {
+      if (!available.includes(b)) continue;
+      const fmt = BACKEND_TO_FORMAT[b];
+      if (!fmt) continue;
+      if (!modelEntry.formats[fmt]) continue;
+      chosen = { backend: b, format: fmt };
+      break;
+    }
+    if (!chosen) {
+      throw new Error(
+        `resolveEngine: no compatible backend found for model ${modelEntry.id}. Available backends: ${available.join(", ")}`
+      );
+    }
+    selectedFormat = chosen.format;
+    selectedBackend = chosen.backend;
+  } else {
+    const fmt = RUNTIME_TO_FORMAT[runtime];
+    if (!fmt) {
+      throw new Error(`resolveEngine: unsupported runtime "${runtime}"`);
+    }
+    if (!modelEntry.formats[fmt]) {
+      throw new Error(
+        `resolveEngine: model ${modelEntry.id} has no ${fmt} format for runtime ${runtime}`
+      );
     }
+    selectedFormat = fmt;
+    selectedBackend = runtime === "onnx" ? backend || "cpu" : runtime;
   }
-});
+  let modelPath;
+  if (models) {
+    modelPath = await models.ensure(modelEntry.id, selectedFormat);
+  } else {
+    modelPath = modelFilePath(modelsDir, modelEntry, selectedFormat);
+    if (!modelExists(modelPath)) {
+      throw new Error(
+        `resolveEngine: model file not found at ${modelPath} and no model service provided`
+      );
+    }
+  }
+  if (selectedFormat === "onnx") {
+    const engine = new NodeInferenceEngine(modelPath, selectedBackend);
+    await engine.initialize();
+    return { engine, format: selectedFormat, modelPath };
+  }
+  const { pythonPath } = options;
+  const PYTHON_SCRIPT_MAP = {
+    coreml: "coreml_inference.py",
+    pytorch: "pytorch_inference.py",
+    openvino: "openvino_inference.py"
+  };
+  const effectiveRuntime = runtime === "auto" ? selectedBackend : runtime;
+  const scriptName = PYTHON_SCRIPT_MAP[effectiveRuntime];
+  if (scriptName && pythonPath) {
+    const candidates = [
+      path2.join(__dirname, "../../python", scriptName),
+      path2.join(__dirname, "../python", scriptName),
+      path2.join(__dirname, "../../../python", scriptName)
+    ];
+    const scriptPath = candidates.find((p) => fs.existsSync(p));
+    if (!scriptPath) {
+      throw new Error(
+        `resolveEngine: Python script "${scriptName}" not found. Searched:
+${candidates.join("\n")}`
+      );
+    }
+    const inputSize = Math.max(modelEntry.inputSize.width, modelEntry.inputSize.height);
+    const engine = new PythonInferenceEngine(pythonPath, scriptPath, effectiveRuntime, modelPath, [
+      `--input-size=${inputSize}`,
+      `--confidence=0.25`
+    ]);
+    await engine.initialize();
+    return { engine, format: selectedFormat, modelPath };
+  }
+  const fallbackPath = modelFilePath(modelsDir, modelEntry, "onnx");
+  if (modelEntry.formats["onnx"] && modelExists(fallbackPath)) {
+    const engine = new NodeInferenceEngine(fallbackPath, "cpu");
+    await engine.initialize();
+    return { engine, format: "onnx", modelPath: fallbackPath };
+  }
+  throw new Error(
+    `resolveEngine: format ${selectedFormat} is not yet supported by NodeInferenceEngine, no Python runtime is available, and no ONNX fallback exists`
+  );
+}
+async function probeOnnxBackends() {
+  const available = ["cpu"];
+  try {
+    const ort = await import("onnxruntime-node");
+    const providers = ort.env?.webgl?.disabled !== void 0 ? ort.InferenceSession?.getAvailableProviders?.() ?? [] : [];
+    for (const p of providers) {
+      const normalized = p.toLowerCase().replace("executionprovider", "");
+      if (normalized === "coreml") available.push("coreml");
+      else if (normalized === "cuda") available.push("cuda");
+      else if (normalized === "tensorrt") available.push("tensorrt");
+    }
+  } catch {
+  }
+  if (process.platform === "darwin" && !available.includes("coreml")) {
+    available.push("coreml");
+  }
+  return [...new Set(available)];
+}
 
 // src/addons/object-detection/index.ts
-var object_detection_exports = {};
-__export(object_detection_exports, {
-  default: () => ObjectDetectionAddon
-});
-module.exports = __toCommonJS(object_detection_exports);
-var import_object_detection_models = __toESM(require_object_detection_models());
-var import_segmentation_models = __toESM(require_segmentation_models());
-var import_types = require("@camstack/types");
-var import_image_utils = __toESM(require_image_utils());
-var import_yolo = __toESM(require_yolo());
-var import_yolo_seg = __toESM(require_yolo_seg());
-var import_engine_resolver = __toESM(require_engine_resolver());
 function isSegModel(modelId) {
   return modelId.includes("-seg");
 }
 var ALL_DETECTION_MODELS = [
-  ...import_object_detection_models.OBJECT_DETECTION_MODELS,
-  ...import_segmentation_models.SEGMENTATION_MODELS
+  ...OBJECT_DETECTION_MODELS,
+  ...SEGMENTATION_MODELS
 ];
 function applyClassMap(detections, classMap) {
   return detections.filter((d) => classMap.mapping[d.class] !== void 0).map((d) => ({
@@ -1489,7 +1284,7 @@ var ObjectDetectionAddon = class {
       const result = await this.engine.runJpeg(frame.data);
       const rawDets = result.detections ?? [];
       const detections2 = rawDets.map((d) => ({
-        class: this.classMapMode === "all" ? d.className : import_types.COCO_TO_MACRO.mapping[d.className] ?? d.className,
+        class: this.classMapMode === "all" ? d.className : import_types3.COCO_TO_MACRO.mapping[d.className] ?? d.className,
         originalClass: d.className,
         score: d.score,
         bbox: {
@@ -1498,7 +1293,7 @@ var ObjectDetectionAddon = class {
           w: (d.bbox[2] - d.bbox[0]) * frame.width,
           h: (d.bbox[3] - d.bbox[1]) * frame.height
         }
-      })).filter((d) => this.classMapMode === "all" || import_types.COCO_TO_MACRO.mapping[d.originalClass] !== void 0);
+      })).filter((d) => this.classMapMode === "all" || import_types3.COCO_TO_MACRO.mapping[d.originalClass] !== void 0);
       return {
         detections: detections2,
         inferenceMs: result.inferenceMs ?? Date.now() - start,
@@ -1507,7 +1302,7 @@ var ObjectDetectionAddon = class {
     }
     const { width: inputW, height: inputH } = this.modelEntry.inputSize;
     const targetSize = Math.max(inputW, inputH);
-    const lb = await (0, import_image_utils.letterbox)(frame.data, targetSize);
+    const lb = await letterbox(frame.data, targetSize);
     const numClasses = this.modelEntry.labels.length;
     const labels = this.modelEntry.labels.map((l) => l.id);
     const postprocessOpts = {
@@ -1535,7 +1330,7 @@ var ObjectDetectionAddon = class {
       const numBoxes = detectionOutput.length / (4 + numClasses + numMaskCoeffs);
       const maskHeight = 160;
       const maskWidth = 160;
-      rawDetections = (0, import_yolo_seg.yoloSegPostprocess)(
+      rawDetections = yoloSegPostprocess(
         {
           detectionOutput,
           protoOutput,
@@ -1550,9 +1345,9 @@ var ObjectDetectionAddon = class {
     } else {
       const output = await this.engine.run(lb.data, [1, 3, targetSize, targetSize]);
       const numBoxes = output.length / (4 + numClasses);
-      rawDetections = (0, import_yolo.yoloPostprocess)(output, numClasses, numBoxes, postprocessOpts);
+      rawDetections = yoloPostprocess(output, numClasses, numBoxes, postprocessOpts);
     }
-    const detections = this.classMapMode === "all" ? rawDetections : applyClassMap(rawDetections, import_types.COCO_TO_MACRO);
+    const detections = this.classMapMode === "all" ? rawDetections : applyClassMap(rawDetections, import_types3.COCO_TO_MACRO);
     return {
       detections,
       inferenceMs: Date.now() - start,
@@ -1583,7 +1378,7 @@ var ObjectDetectionAddon = class {
         }
       }
     }
-    const resolved = await (0, import_engine_resolver.resolveEngine)({
+    const resolved = await resolveEngine({
       runtime,
       backend,
       modelEntry: entry,
@@ -1693,7 +1488,7 @@ var ObjectDetectionAddon = class {
     };
   }
   getClassMap() {
-    return import_types.COCO_TO_MACRO;
+    return import_types3.COCO_TO_MACRO;
   }
   getModelCatalog() {
     return [...ALL_DETECTION_MODELS];
@@ -1702,7 +1497,7 @@ var ObjectDetectionAddon = class {
     return [];
   }
   getActiveLabels() {
-    return this.classMapMode === "all" ? import_types.COCO_80_LABELS : import_types.MACRO_LABELS;
+    return this.classMapMode === "all" ? import_types3.COCO_80_LABELS : import_types3.MACRO_LABELS;
   }
   async probe() {
     return {