suryajayaraman / sort_resources

References to implementation of SORT algorithm

SORT_resources

• C++ implementation of SORT algorithm
• SORT (Simple and online realtime tracking) is an online tracking algorithm where only detections from the previous and the current frame are presented to the tracker.
• Kalman Filter + Hungarian algorithm integral for the algorithm

• The repo uses Eigen library for matrix operations and hungarian algorithm implementation
• sort.py from official repo is the reference python implementation for this repo. Check the original repo on how to run the python script.

Reference Folder Structure

projectRoot/

- data/
    train/
        ADL-Rundle-6/
            det/det/txt
        ADL-Rundle-8/
            det/det.txt
        .....

- mot_benchmark
    train/
        ADL-Rundle-6/
            det/det/txt
            gt/gt.txt
            img1/
                000001.jpg
                000002.jpg
                ......

• The data/ folder contains necessary bounding box information for all frames in all scenarios
• The mot_benchmark/ folder contains corresponding camera frames and is optional (~1.5GB download size)

Requirements

The code was developed on Ubuntu 20.04 LTS OS with

• g++ 9.4.0
• cmake 3.16.3
• Eigen 3.3
• OpenCv

python script runtime speed

References

• SORT original paper
• SORT official github implementation
• SORT cpp implementation
• SORT python implementation with importance to reidentification