Zhongdao/Towards-Realtime-MOT
A real-time multi-object tracking system using a joint detection and embedding neural network architecture.
Not currently ranked — collecting fresh signals.
star history
This repository implements the Joint Detection and Embedding (JDE) model for fast multi-object tracking. The system uses a shared neural network to simultaneously learn object detection and appearance embedding for tracking-by-detection. Built with PyTorch, it achieves near real-time performance (22-38 FPS) and is published at ECCV 2020.
Frequently asked
- What is Zhongdao/Towards-Realtime-MOT?
- A real-time multi-object tracking system using a joint detection and embedding neural network architecture.
- Is Towards-Realtime-MOT open source?
- Yes — Zhongdao/Towards-Realtime-MOT is open source, released under the MIT license.
- What language is Towards-Realtime-MOT written in?
- Zhongdao/Towards-Realtime-MOT is primarily written in Python.
- How popular is Towards-Realtime-MOT?
- Zhongdao/Towards-Realtime-MOT has 2.4k stars on GitHub.
- Where can I find Towards-Realtime-MOT?
- Zhongdao/Towards-Realtime-MOT is on GitHub at https://github.com/Zhongdao/Towards-Realtime-MOT.