Introduction

Java implementation of SORT: Simple, online, and real-time tracking of multiple objects in a video sequence. SORT was initially described in this paper. At the time of the initial publication, SORT was ranked the best open source multiple object tracker on the MOT benchmark.

The original SORT python implementation and C++ implementation can be found here.

Dependencies

• Ubuntu 18.04
• OpenCV 4.5.4
• commons-io-2.11.0

Demo:

The code is run using IntelliJ

To run the tracker with the provided detections and visualize the results:

1. Download the 2D MOT 2015 benchmark dataset

2. Create a symbolic link to the dataset

   $ ln -s /path/to/MOT2015_challenge/data/2DMOT2015 /path/to/sort-cpp/mot_benchmark

3. Run the demo

   To enable visualize, set display=True

Evaluate Metrics

Using the Python implementation of metrics for benchmarking multiple object trackers (MOT) to evaluate metrics.

Dataset Structure

Layout for ground truth data
    <GT_ROOT>/<SEQUENCE_1>/gt/gt.txt
    <GT_ROOT>/<SEQUENCE_2>/gt/gt.txt
    ...

Layout for test data
    <TEST_ROOT>/<SEQUENCE_1>.txt
    <TEST_ROOT>/<SEQUENCE_2>.txt
    ...

Example:
mot_benchmark
├── test
│   ├── ADL-Rundle-6.txt
│   └── ADL-Rundle-8.txt
└── train
    ├── ADL-Rundle-6
    │   └── gt
    │       └── gt.txt
    └── ADL-Rundle-8
        └── gt
            └── gt.txt


Sequences of ground truth and test will be matched according to the `<SEQUENCE_X>`
string.

Example

# Optional for virtualenv
$ source ~/env/bin/activate
$ pip install motmetrics
# Usage
$ python -m motmetrics.apps.eval_motchallenge --help
# Format: python -m motmetrics.apps.eval_motchallenge groundtruths tests
$ python -m motmetrics.apps.eval_motchallenge mot_benchmark/train output/

Result

References

1. https://github.com/abewley/sort
2. https://github.com/mcximing/sort-cpp