This is the official repository for our paper, "Multiple Toddler Tracking in Indoor Videos" accpted for publication in CV4Smalls Workshop at WACV 2024.

MTTSort: Enhanced DeepSort for Toddler Tracking

MTTSort is an enhanced version of the DeepSort algorithm, specifically tailored for tracking toddlers in various environments. By leveraging advanced deep learning techniques, MTTSort provides accurate and real-time tracking, making it an ideal solution for monitoring and ensuring the safety of toddlers.

Features

• Real-time toddler tracking with high accuracy.
• Advanced deep learning techniques for improved performance.
• Customizable tracking parameters for different environments.
• Integration with YOLOv8 for robust object detection.
• Genetic Algorithm Evolution for fine-grained hyper-parameters tuning.

Getting Started

Follow these steps to set up and run MTTSort on your local machine.

Prerequisites

Before you begin, ensure you have met the following requirements:

• Python 3.8 or higher
• Pip package manager

Installation

1. Clone the repository:

   git clone https://github.com/your-username/MTTSort.git

2. Navigate to the cloned directory:

   cd MTTSort

3. Install the dependencies:

   pip install -r requirements.txt

Usage

To use MTTSort for toddler tracking, follow these steps:

1. Set the working directory:

   cd path/to/detect/folder

2. Run the genetic algorithm script for evolution:

   python evolve.py

3. Run the detection model with HOTA and MOTA and IDF1 scores calculation:

   python predict_evaluate.py

4. Run only model for detections and tracks:

   python predict_frames.py

Data Preperation

For annotation and labeling of images used in testing the MTTSort model, LabelMe was utilized. LabelMe is an open-source graphical image annotation tool written in Python.

After preparing the JSON files from Labelme run the json_to_text.py script to transform the labels to the desired format.

1. Run to transform the json labels to text:

   python json_to_text.py

Evaluation

The data used for evaluation should be in such format

    frame_id, Subject_ID, x_top, y_top, x_bottom, y_bottom

Results

Here's a demonstration of the MTTSort algorithm in action:

### Support and Contributions

For support, questions, or contributions, please open an issue or submit a pull request in the repository.

Citation

If you find use this code or dataset for your research, please consider citing our paper:

@inproceedings{Amraee2024toddlertracking,
  title={Multiple Toddler Tracking in Indoor Videos},
  author={Amraee, Somaieh and Galoaa, Bishoy  and Goodwin, Matthew  and  Hatamimajoumerd,  Elaheh  and Ostadabbas, Sarah},
  booktitle={IEEE/CVF Winter Conference on Applications of Computer Vision Workshops (WACVW)},
  month={1},
  year={2024}
  }

License

This project is licensed under the [Your License Name]. For more information, see the LICENSE file in the repository.

Acknowledgements

Special thanks to the original DeepSort and YOLOv8 contributors for their foundational work.