Search tools, frameworks, libraries, categories and domains
PyTorch's official package of vision datasets, transforms and pretrained models.
pip install torchvision
Package torchvision, last 26 weeks.
torchvision
Other tools in the same categories.
opencv/opencv
The classic computer vision library for image processing, video I/O and camera pipelines.
ultralytics/ultralytics
Most-used real-time object detection, segmentation and pose library; train and deploy YOLO models easily.
facebookresearch/sam3
Meta's Segment Anything Model 3: promptable detection, segmentation and tracking from text or clicks.
roboflow/supervision
Roboflow's reusable utilities for detection post-processing, tracking, zones and annotation.
roboflow/rf-detr
Roboflow's real-time transformer detector that beats YOLO-class models on accuracy at similar speed.
huggingface/pytorch-image-models
PyTorch Image Models: hundreds of pretrained vision backbones with training and eval scripts.
kornia/kornia
Differentiable computer vision library for PyTorch: geometry, augmentation and feature matching.
facebookresearch/detectron2
Meta's detection and segmentation research platform; still cited widely but rarely updated.
cvat-ai/cvat
CVAT is a leading data annotation platform for image, video, audio, and 3D datasets. It offers open-source, cloud, and enterprise…