xingyizhou/CenterTrack

Health v2 · maintenance only

Flags: no_releases

Adoption not part of the score

2478 stars · 515 forks observed · 2026-08-28

What it is AI-extracted, prompt v1, taxonomy v1, 2026-08-30, confidence not recorded

CenterTrack is a deep learning model and research codebase that performs simultaneous multi-object detection and tracking using center points, as published at ECCV 2020. It supports 2D tracking, monocular 3D tracking, multi-category tracking, and pose tracking, with state-of-the-art results on MOT17, KITTI, and nuScenes benchmarks.

Use cases

• track pedestrians in video in real time
• detect and track vehicles for autonomous driving research
• run monocular 3D object tracking on nuScenes
• train a tracker on still image datasets without video
• reproduce MOT17 and KITTI tracking benchmark results
• extend tracking to multi-category or pose tracking

When to choose

• you need an online, real-time joint detection-and-tracking model
• you want a research baseline for multi-object or 3D tracking benchmarks
• you only have static image datasets for training

When to avoid

• you need a production-ready maintained pipeline rather than research code
• you need offline or batch tracking with future-frame information
• you work outside Python/PyTorch environments

Facets

library · maturity maintenance

computer-vision machine-learning deep-learning image-processing computer-vision machine-learning autonomous-vehicles python object-detection object-tracking multi-object-tracking 3d-tracking centerpoint research-code pytorch linux gpu

Member repositories

For agents

markdown · JSON · MCP: product_card(name="xingyizhou/CenterTrack")