visionml/pytracking

Health v2 · maintenance only

Adoption not part of the score

3514 stars · 614 forks observed · 2026-08-28

What it is AI-extracted, prompt v1, taxonomy v1, 2026-08-29, confidence not recorded

PyTracking is a PyTorch-based framework for visual object tracking and video object segmentation, providing official implementations of trackers like ATOM, DiMP, PrDiMP, KYS, KeepTrack, ToMP, RTS, and TaMOs. It includes evaluation libraries, a training framework (LTR), and a model zoo with pretrained models.

Use cases

• track objects in videos with deep learning
• train a custom visual tracker
• evaluate trackers on standard tracking benchmarks
• run multi-object tracking with TaMOs
• perform video object segmentation
• reproduce results from tracking research papers

When to choose

• you need state-of-the-art single or multi-object tracking implementations in PyTorch
• you want to train tracking networks with a dedicated framework
• you need benchmark evaluation on standard tracking datasets

When to avoid

• you need lightweight real-time tracking on edge devices without GPU
• you want a plug-and-play tracker for production apps rather than a research framework
• your project requires a permissive license (GPL-3.0)

Facets

library · maturity active

machine-learning computer-vision image-processing benchmarking testing computer-vision machine-learning deep-learning python visual-tracking object-tracking video-object-segmentation pytorch research model-zoo multi-object-tracking video linux gpu

Member repositories

For agents

markdown · JSON · MCP: product_card(name="visionml/pytracking")