Awesome

Official MOTChallenge Evaluation Kit

This repository contains the evaluation scripts for all challenges, available at www.MOTChallenge.net. This devkit replaces the previous version that used to be accessible over https://bitbucket.org/amilan/motchallenge-devkit and is no longer maintained.

- IMPORTANT!
- The MOT evaluation code is not any longer maintained. 

Please visit the new official python evaluation code.

Requirements

• Python (3.6.9 or newer)
• MATLAB (tested with R2020a, other versions should work too)

Directories

All evaluation scripts assume the following directory structure:

./res

This directory contains the tracking results (for each sequence); result files should be placed to subfolders

./seqmaps

Sequence lists for all supported different benchmarks

./data

This directory contains the ground truth data (for several different sequences/challenges)

./vid

This directory is for the visualization of results or annotations

Evaluation Scripts

This repo provides the evaluation scripts for the following challenges of www.motchallenge.net:

MOT - Multi-Object Tracking - MOT15, MOT16, MOT17, MOT20

- IMPORTANT!
- The MOT evaluation code is not any longer maintained. 

Please visit the new official python evaluation code.

MOT Evaluation

DET - Multi-Object Detection - MOT16Det, MOT20Det

DET Evaluation

MOTS - Multi-Object Tracking and Segmentation - MOTS20

MOTS Evaluation

3D-ZeF20

ZF3D Evaluation

TAO - Tracking Any Object Challenge

TAO Evaluation

Feedback and contact

We are constantly working on improving our benchmark to provide the best performance to the community. You can help us to make the benchmark better to open issues in the repo and report bugs to the author:

Patrick Dendorfer - patrick.dendorfer@tum.de