@huggingface/tasks 0.13.0 → 0.13.1-test

package/src/tasks/fill-mask/spec/input.json

@@ -1,38 +0,0 @@
-{
-	"$id": "/inference/schemas/fill-mask/input.json",
-	"$schema": "http://json-schema.org/draft-06/schema#",
-	"description": "Inputs for Fill Mask inference",
-	"title": "FillMaskInput",
-	"type": "object",
-	"properties": {
-		"inputs": {
-			"description": "The text with masked tokens",
-			"type": "string"
-		},
-		"parameters": {
-			"description": "Additional inference parameters",
-			"$ref": "#/$defs/FillMaskParameters"
-		}
-	},
-	"$defs": {
-		"FillMaskParameters": {
-			"title": "FillMaskParameters",
-			"description": "Additional inference parameters for Fill Mask",
-			"type": "object",
-			"properties": {
-				"top_k": {
-					"type": "integer",
-					"description": "When passed, overrides the number of predictions to return."
-				},
-				"targets": {
-					"description": "When passed, the model will limit the scores to the passed targets instead of looking up in the whole vocabulary. If the provided targets are not in the model vocab, they will be tokenized and the first resulting token will be used (with a warning, and that might be slower).",
-					"type": "array",
-					"items": {
-						"type": "string"
-					}
-				}
-			}
-		}
-	},
-	"required": ["inputs"]
-}

package/src/tasks/fill-mask/spec/output.json

@@ -1,29 +0,0 @@
-{
-	"$id": "/inference/schemas/fill-mask/output.json",
-	"$schema": "http://json-schema.org/draft-06/schema#",
-	"description": "Outputs of inference for the Fill Mask task",
-	"title": "FillMaskOutput",
-	"type": "array",
-	"items": {
-		"type": "object",
-		"properties": {
-			"sequence": {
-				"type": "string",
-				"description": "The corresponding input with the mask token prediction."
-			},
-			"score": {
-				"type": "number",
-				"description": "The corresponding probability"
-			},
-			"token": {
-				"type": "integer",
-				"description": "The predicted token id (to replace the masked one)."
-			},
-			"token_str": {
-				"type": "string",
-				"description": "The predicted token (to replace the masked one)."
-			}
-		},
-		"required": ["sequence", "score", "token", "tokenStr"]
-	}
-}

package/src/tasks/image-classification/about.md

@@ -1,50 +0,0 @@
-## Use Cases
-
-Image classification models can be used when we are not interested in specific instances of objects with location information or their shape.
-
-### Keyword Classification
-
-Image classification models are used widely in stock photography to assign each image a keyword.
-
-### Image Search
-
-Models trained in image classification can improve user experience by organizing and categorizing photo galleries on the phone or in the cloud, on multiple keywords or tags.
-
-## Inference
-
-With the `transformers` library, you can use the `image-classification` pipeline to infer with image classification models. You can initialize the pipeline with a model id from the Hub. If you do not provide a model id it will initialize with [google/vit-base-patch16-224](https://huggingface.co/google/vit-base-patch16-224) by default. When calling the pipeline you just need to specify a path, http link or an image loaded in PIL. You can also provide a `top_k` parameter which determines how many results it should return.
-
-```python
-from transformers import pipeline
-clf = pipeline("image-classification")
-clf("path_to_a_cat_image")
-
-[{'label': 'tabby cat', 'score': 0.731},
-...
-]
-```
-
-You can use [huggingface.js](https://github.com/huggingface/huggingface.js) to classify images using models on Hugging Face Hub.
-
-```javascript
-import { HfInference } from "@huggingface/inference";
-
-const inference = new HfInference(HF_TOKEN);
-await inference.imageClassification({
-	data: await (await fetch("https://picsum.photos/300/300")).blob(),
-	model: "microsoft/resnet-50",
-});
-```
-
-## Useful Resources
-
-- [Let's Play Pictionary with Machine Learning!](https://www.youtube.com/watch?v=LS9Y2wDVI0k)
-- [Fine-Tune ViT for Image Classification with 🤗Transformers](https://huggingface.co/blog/fine-tune-vit)
-- [Walkthrough of Computer Vision Ecosystem in Hugging Face - CV Study Group](https://www.youtube.com/watch?v=oL-xmufhZM8)
-- [Computer Vision Study Group: Swin Transformer](https://www.youtube.com/watch?v=Ngikt-K1Ecc)
-- [Computer Vision Study Group: Masked Autoencoders Paper Walkthrough](https://www.youtube.com/watch?v=Ngikt-K1Ecc)
-- [Image classification task guide](https://huggingface.co/docs/transformers/tasks/image_classification)
-
-### Creating your own image classifier in just a few minutes
-
-With [HuggingPics](https://github.com/nateraw/huggingpics), you can fine-tune Vision Transformers for anything using images found on the web. This project downloads images of classes defined by you, trains a model, and pushes it to the Hub. You even get to try out the model directly with a working widget in the browser, ready to be shared with all your friends!

package/src/tasks/image-classification/data.ts

@@ -1,88 +0,0 @@
-import type { TaskDataCustom } from "..";
-
-const taskData: TaskDataCustom = {
-	datasets: [
-		{
-			// TODO write proper description
-			description: "Benchmark dataset used for image classification with images that belong to 100 classes.",
-			id: "cifar100",
-		},
-		{
-			// TODO write proper description
-			description: "Dataset consisting of images of garments.",
-			id: "fashion_mnist",
-		},
-	],
-	demo: {
-		inputs: [
-			{
-				filename: "image-classification-input.jpeg",
-				type: "img",
-			},
-		],
-		outputs: [
-			{
-				type: "chart",
-				data: [
-					{
-						label: "Egyptian cat",
-						score: 0.514,
-					},
-					{
-						label: "Tabby cat",
-						score: 0.193,
-					},
-					{
-						label: "Tiger cat",
-						score: 0.068,
-					},
-				],
-			},
-		],
-	},
-	metrics: [
-		{
-			description: "",
-			id: "accuracy",
-		},
-		{
-			description: "",
-			id: "recall",
-		},
-		{
-			description: "",
-			id: "precision",
-		},
-		{
-			description: "",
-			id: "f1",
-		},
-	],
-	models: [
-		{
-			description: "A strong image classification model.",
-			id: "google/vit-base-patch16-224",
-		},
-		{
-			description: "A robust image classification model.",
-			id: "facebook/deit-base-distilled-patch16-224",
-		},
-		{
-			description: "A strong image classification model.",
-			id: "facebook/convnext-large-224",
-		},
-	],
-	spaces: [
-		{
-			// TO DO: write description
-			description: "An application that classifies what a given image is about.",
-			id: "nielsr/perceiver-image-classification",
-		},
-	],
-	summary:
-		"Image classification is the task of assigning a label or class to an entire image. Images are expected to have only one class for each image. Image classification models take an image as input and return a prediction about which class the image belongs to.",
-	widgetModels: ["google/vit-base-patch16-224"],
-	youtubeId: "tjAIM7BOYhw",
-};
-
-export default taskData;

package/src/tasks/image-classification/inference.ts

@@ -1,52 +0,0 @@
-/**
- * Inference code generated from the JSON schema spec in ./spec
- *
- * Using src/scripts/inference-codegen
- */
-/**
- * Inputs for Image Classification inference
- */
-export interface ImageClassificationInput {
-	/**
-	 * The input image data as a base64-encoded string. If no `parameters` are provided, you can
-	 * also provide the image data as a raw bytes payload.
-	 */
-	inputs: string;
-	/**
-	 * Additional inference parameters
-	 */
-	parameters?: ImageClassificationParameters;
-	[property: string]: unknown;
-}
-/**
- * Additional inference parameters
- *
- * Additional inference parameters for Image Classification
- */
-export interface ImageClassificationParameters {
-	function_to_apply?: ClassificationOutputTransform;
-	/**
-	 * When specified, limits the output to the top K most probable classes.
-	 */
-	top_k?: number;
-	[property: string]: unknown;
-}
-/**
- * The function to apply to the model outputs in order to retrieve the scores.
- */
-export type ClassificationOutputTransform = "sigmoid" | "softmax" | "none";
-export type ImageClassificationOutput = ImageClassificationOutputElement[];
-/**
- * Outputs of inference for the Image Classification task
- */
-export interface ImageClassificationOutputElement {
-	/**
-	 * The predicted class label.
-	 */
-	label: string;
-	/**
-	 * The corresponding probability.
-	 */
-	score: number;
-	[property: string]: unknown;
-}

package/src/tasks/image-classification/spec/input.json

@@ -1,35 +0,0 @@
-{
-	"$id": "/inference/schemas/image-classification/input.json",
-	"$schema": "http://json-schema.org/draft-06/schema#",
-	"description": "Inputs for Image Classification inference",
-	"title": "ImageClassificationInput",
-	"type": "object",
-	"properties": {
-		"inputs": {
-			"type": "string",
-			"description": "The input image data as a base64-encoded string. If no `parameters` are provided, you can also provide the image data as a raw bytes payload."
-		},
-		"parameters": {
-			"description": "Additional inference parameters",
-			"$ref": "#/$defs/ImageClassificationParameters"
-		}
-	},
-	"$defs": {
-		"ImageClassificationParameters": {
-			"title": "ImageClassificationParameters",
-			"description": "Additional inference parameters for Image Classification",
-			"type": "object",
-			"properties": {
-				"function_to_apply": {
-					"title": "ImageClassificationOutputTransform",
-					"$ref": "/inference/schemas/common-definitions.json#/definitions/ClassificationOutputTransform"
-				},
-				"top_k": {
-					"type": "integer",
-					"description": "When specified, limits the output to the top K most probable classes."
-				}
-			}
-		}
-	},
-	"required": ["inputs"]
-}

package/src/tasks/image-classification/spec/output.json

@@ -1,11 +0,0 @@
-{
-	"$id": "/inference/schemas/image-classification/output.json",
-	"$schema": "http://json-schema.org/draft-06/schema#",
-	"description": "Outputs of inference for the Image Classification task",
-	"title": "ImageClassificationOutput",
-	"type": "array",
-	"items": {
-		"type": "object",
-		"$ref": "/inference/schemas/common-definitions.json#/definitions/ClassificationOutput"
-	}
-}

package/src/tasks/image-feature-extraction/about.md

@@ -1,23 +0,0 @@
-## Use Cases
-
-### Transfer Learning
-
-Models trained on a specific dataset can learn features about the data. For instance, a model trained on a car classification dataset learns to recognize edges and curves on a very high level and car-specific features on a low level. This information can be transferred to a new model that is going to be trained on classifying trucks. This process of extracting features and transferring to another model is called transfer learning.
-
-### Similarity
-
-Features extracted from models contain semantically meaningful information about the world. These features can be used to detect the similarity between two images. Assume there are two images: a photo of a stray cat in a street setting and a photo of a cat at home. These images both contain cats, and the features will contain the information that there's a cat in the image. Thus, comparing the features of a stray cat photo to the features of a domestic cat photo will result in higher similarity compared to any other image that doesn't contain any cats.
-
-## Inference
-
-```python
-import torch
-from transformers import pipeline
-
-pipe = pipeline(task="image-feature-extraction", model_name="google/vit-base-patch16-384", framework="pt", pool=True)
-pipe("https://huggingface.co/datasets/huggingface/documentation-images/resolve/main/cats.png")
-
-feature_extractor(text,return_tensors = "pt")[0].numpy().mean(axis=0)
-
-'[[[0.21236686408519745, 1.0919708013534546, 0.8512550592422485, ...]]]'
-```

package/src/tasks/image-feature-extraction/data.ts

@@ -1,59 +0,0 @@
-import type { TaskDataCustom } from "..";
-
-const taskData: TaskDataCustom = {
-	datasets: [
-		{
-			description:
-				"ImageNet-1K is a image classification dataset in which images are used to train image-feature-extraction models.",
-			id: "imagenet-1k",
-		},
-	],
-	demo: {
-		inputs: [
-			{
-				filename: "mask-generation-input.png",
-				type: "img",
-			},
-		],
-		outputs: [
-			{
-				table: [
-					["Dimension 1", "Dimension 2", "Dimension 3"],
-					["0.21236686408519745", "1.0919708013534546", "0.8512550592422485"],
-					["0.809657871723175", "-0.18544459342956543", "-0.7851548194885254"],
-					["1.3103108406066895", "-0.2479034662246704", "-0.9107287526130676"],
-					["1.8536205291748047", "-0.36419737339019775", "0.09717650711536407"],
-				],
-				type: "tabular",
-			},
-		],
-	},
-	metrics: [],
-	models: [
-		{
-			description: "A powerful image feature extraction model.",
-			id: "timm/vit_large_patch14_dinov2.lvd142m",
-		},
-		{
-			description: "A strong image feature extraction model.",
-			id: "nvidia/MambaVision-T-1K",
-		},
-		{
-			description: "A robust image feature extraction model.",
-			id: "facebook/dino-vitb16",
-		},
-		{
-			description: "Strong image feature extraction model made for information retrieval from documents.",
-			id: "vidore/colpali",
-		},
-		{
-			description: "Strong image feature extraction model that can be used on images and documents.",
-			id: "OpenGVLab/InternViT-6B-448px-V1-2",
-		},
-	],
-	spaces: [],
-	summary: "Image feature extraction is the task of extracting features learnt in a computer vision model.",
-	widgetModels: [],
-};
-
-export default taskData;

package/src/tasks/image-segmentation/about.md

@@ -1,63 +0,0 @@
-## Use Cases
-
-### Autonomous Driving
-
-Segmentation models are used to identify road patterns such as lanes and obstacles for safer driving.
-
-### Background Removal
-
-Image Segmentation models are used in cameras to erase the background of certain objects and apply filters to them.
-
-### Medical Imaging
-
-Image Segmentation models are used to distinguish organs or tissues, improving medical imaging workflows. Models are used to segment dental instances, analyze X-Ray scans or even segment cells for pathological diagnosis. This [dataset](https://github.com/v7labs/covid-19-xray-dataset) contains images of lungs of healthy patients and patients with COVID-19 segmented with masks. Another [segmentation dataset](https://ivdm3seg.weebly.com/data.html) contains segmented MRI data of the lower spine to analyze the effect of spaceflight simulation.
-
-## Task Variants
-
-### Semantic Segmentation
-
-Semantic Segmentation is the task of segmenting parts of an image that belong to the same class. Semantic Segmentation models make predictions for each pixel and return the probabilities of the classes for each pixel. These models are evaluated on Mean Intersection Over Union (Mean IoU).
-
-### Instance Segmentation
-
-Instance Segmentation is the variant of Image Segmentation where every distinct object is segmented, instead of one segment per class.
-
-### Panoptic Segmentation
-
-Panoptic Segmentation is the Image Segmentation task that segments the image both by instance and by class, assigning each pixel a different instance of the class.
-
-## Inference
-
-You can infer with Image Segmentation models using the `image-segmentation` pipeline. You need to install [timm](https://github.com/rwightman/pytorch-image-models) first.
-
-```python
-!pip install timm
-model = pipeline("image-segmentation")
-model("cat.png")
-#[{'label': 'cat',
-#  'mask': mask_code,
-#  'score': 0.999}
-# ...]
-```
-
-You can use [huggingface.js](https://github.com/huggingface/huggingface.js) to infer image segmentation models on Hugging Face Hub.
-
-```javascript
-import { HfInference } from "@huggingface/inference";
-
-const inference = new HfInference(HF_TOKEN);
-await inference.imageSegmentation({
-	data: await (await fetch("https://picsum.photos/300/300")).blob(),
-	model: "facebook/mask2former-swin-base-coco-panoptic",
-});
-```
-
-## Useful Resources
-
-Would you like to learn more about image segmentation? Great! Here you can find some curated resources that you may find helpful!
-
-- [Fine-Tune a Semantic Segmentation Model with a Custom Dataset](https://huggingface.co/blog/fine-tune-segformer)
-- [Walkthrough of Computer Vision Ecosystem in Hugging Face - CV Study Group](https://www.youtube.com/watch?v=oL-xmufhZM8)
-- [A Guide on Universal Image Segmentation with Mask2Former and OneFormer](https://huggingface.co/blog/mask2former)
-- [Zero-shot image segmentation with CLIPSeg](https://huggingface.co/blog/clipseg-zero-shot)
-- [Semantic segmentation task guide](https://huggingface.co/docs/transformers/tasks/semantic_segmentation)

package/src/tasks/image-segmentation/data.ts

@@ -1,99 +0,0 @@
-import type { TaskDataCustom } from "..";
-
-const taskData: TaskDataCustom = {
-	datasets: [
-		{
-			description: "Scene segmentation dataset.",
-			id: "scene_parse_150",
-		},
-	],
-	demo: {
-		inputs: [
-			{
-				filename: "image-segmentation-input.jpeg",
-				type: "img",
-			},
-		],
-		outputs: [
-			{
-				filename: "image-segmentation-output.png",
-				type: "img",
-			},
-		],
-	},
-	metrics: [
-		{
-			description:
-				"Average Precision (AP) is the Area Under the PR Curve (AUC-PR). It is calculated for each semantic class separately",
-			id: "Average Precision",
-		},
-		{
-			description: "Mean Average Precision (mAP) is the overall average of the AP values",
-			id: "Mean Average Precision",
-		},
-		{
-			description:
-				"Intersection over Union (IoU) is the overlap of segmentation masks. Mean IoU is the average of the IoU of all semantic classes",
-			id: "Mean Intersection over Union",
-		},
-		{
-			description: "APα is the Average Precision at the IoU threshold of a α value, for example, AP50 and AP75",
-			id: "APα",
-		},
-	],
-	models: [
-		{
-			// TO DO: write description
-			description: "Solid semantic segmentation model trained on ADE20k.",
-			id: "openmmlab/upernet-convnext-small",
-		},
-		{
-			description: "Background removal model.",
-			id: "briaai/RMBG-1.4",
-		},
-		{
-			description: "A multipurpose image segmentation model for high resolution images.",
-			id: "ZhengPeng7/BiRefNet",
-		},
-		{
-			description: "Powerful human-centric image segmentation model.",
-			id: "facebook/sapiens-seg-1b",
-		},
-		{
-			description: "Panoptic segmentation model trained on the COCO (common objects) dataset.",
-			id: "facebook/mask2former-swin-large-coco-panoptic",
-		},
-	],
-	spaces: [
-		{
-			description: "A semantic segmentation application that can predict unseen instances out of the box.",
-			id: "facebook/ov-seg",
-		},
-		{
-			description: "One of the strongest segmentation applications.",
-			id: "jbrinkma/segment-anything",
-		},
-		{
-			description: "A human-centric segmentation model.",
-			id: "facebook/sapiens-pose",
-		},
-		{
-			description: "An instance segmentation application to predict neuronal cell types from microscopy images.",
-			id: "rashmi/sartorius-cell-instance-segmentation",
-		},
-		{
-			description: "An application that segments videos.",
-			id: "ArtGAN/Segment-Anything-Video",
-		},
-		{
-			description: "An panoptic segmentation application built for outdoor environments.",
-			id: "segments/panoptic-segment-anything",
-		},
-	],
-	summary:
-		"Image Segmentation divides an image into segments where each pixel in the image is mapped to an object. This task has multiple variants such as instance segmentation, panoptic segmentation and semantic segmentation.",
-	widgetModels: ["nvidia/segformer-b0-finetuned-ade-512-512"],
-	youtubeId: "dKE8SIt9C-w",
-};
-
-export default taskData;

package/src/tasks/image-segmentation/inference.ts

@@ -1,69 +0,0 @@
-/**
- * Inference code generated from the JSON schema spec in ./spec
- *
- * Using src/scripts/inference-codegen
- */
-/**
- * Inputs for Image Segmentation inference
- */
-export interface ImageSegmentationInput {
-	/**
-	 * The input image data as a base64-encoded string. If no `parameters` are provided, you can
-	 * also provide the image data as a raw bytes payload.
-	 */
-	inputs: string;
-	/**
-	 * Additional inference parameters
-	 */
-	parameters?: ImageSegmentationParameters;
-	[property: string]: unknown;
-}
-/**
- * Additional inference parameters
- *
- * Additional inference parameters for Image Segmentation
- */
-export interface ImageSegmentationParameters {
-	/**
-	 * Threshold to use when turning the predicted masks into binary values.
-	 */
-	mask_threshold?: number;
-	/**
-	 * Mask overlap threshold to eliminate small, disconnected segments.
-	 */
-	overlap_mask_area_threshold?: number;
-	/**
-	 * Segmentation task to be performed, depending on model capabilities.
-	 */
-	subtask?: ImageSegmentationSubtask;
-	/**
-	 * Probability threshold to filter out predicted masks.
-	 */
-	threshold?: number;
-	[property: string]: unknown;
-}
-/**
- * Segmentation task to be performed, depending on model capabilities.
- */
-export type ImageSegmentationSubtask = "instance" | "panoptic" | "semantic";
-export type ImageSegmentationOutput = ImageSegmentationOutputElement[];
-/**
- * Outputs of inference for the Image Segmentation task
- *
- * A predicted mask / segment
- */
-export interface ImageSegmentationOutputElement {
-	/**
-	 * The label of the predicted segment.
-	 */
-	label: string;
-	/**
-	 * The corresponding mask as a black-and-white image (base64-encoded).
-	 */
-	mask: string;
-	/**
-	 * The score or confidence degree the model has.
-	 */
-	score?: number;
-	[property: string]: unknown;
-}

package/src/tasks/image-segmentation/spec/input.json

@@ -1,45 +0,0 @@
-{
-	"$id": "/inference/schemas/image-segmentation/input.json",
-	"$schema": "http://json-schema.org/draft-06/schema#",
-	"description": "Inputs for Image Segmentation inference",
-	"title": "ImageSegmentationInput",
-	"type": "object",
-	"properties": {
-		"inputs": {
-			"type": "string",
-			"description": "The input image data as a base64-encoded string. If no `parameters` are provided, you can also provide the image data as a raw bytes payload."
-		},
-		"parameters": {
-			"description": "Additional inference parameters",
-			"$ref": "#/$defs/ImageSegmentationParameters"
-		}
-	},
-	"$defs": {
-		"ImageSegmentationParameters": {
-			"title": "ImageSegmentationParameters",
-			"description": "Additional inference parameters for Image Segmentation",
-			"type": "object",
-			"properties": {
-				"mask_threshold": {
-					"type": "number",
-					"description": "Threshold to use when turning the predicted masks into binary values."
-				},
-				"overlap_mask_area_threshold": {
-					"type": "number",
-					"description": "Mask overlap threshold to eliminate small, disconnected segments."
-				},
-				"subtask": {
-					"title": "ImageSegmentationSubtask",
-					"type": "string",
-					"description": "Segmentation task to be performed, depending on model capabilities.",
-					"enum": ["instance", "panoptic", "semantic"]
-				},
-				"threshold": {
-					"type": "number",
-					"description": "Probability threshold to filter out predicted masks."
-				}
-			}
-		}
-	},
-	"required": ["inputs"]
-}