A multi-camera multi-person tracking system using YOLOv8 and DeepOCSORT. Includes additional features such as emotion recognition, gender recognition, and posture classification.

1. Create a conda environment

   conda create -n mct python=3.8.17
   conda activate mct

2. Clone this repository and install necessary dependencies

   pip install -r requirements.txt 

3. To use CUDA, install onnxruntime-gpu based on your CUDA version as shown here: https://onnxruntime.ai/docs/execution-providers/CUDA-ExecutionProvider.html

• Input videos: To specify video sources, navigate to the configs.yaml file and locate the sources field. In this section, you can define multiple video sources that your application will process. (Note: videos will be processed in parallel)

  Example:

  sources:
  - /path/to/video1.mp4
  - /path/to/video2.mp4
  - /path/to/video3.mp4

• Output Location: If you wish to customize the output location for the processed data, find the output_location entry within the configs.yaml file.

  Example:

  output_location: /path/to/output/folder

Within the configs.yaml file, you have the flexibility to adjust essential hyperparameters that influence the behavior of your application. Three key hyperparameters, namely tracker_configs, person_seg_configs, and face_det_configs, are defined in this configuration file.

• Tracker Configuration: The tracker_configs section allows you to fine-tune parameters associated with the tracking module. Depending on your specific requirements, you can experiment with values such as reid_thresh, assoc_func, iou_thresh, etc.

• Person Segmentation and Face Detection Configuration: The person_seg_configs and face_det_configs sections correspond to modules trained using Ultralytics. Refer to their documentation for detailed parameter information.

You can choose to anonymize the people by setting anonymize to true in configs.yaml. Enabling the this parameter ensures that each detected person is anonymized by applying a black mask.

Once you have completed the setup process and configured your parameters as desired, run the script:

python main.py --configs configs.yaml

• https://github.com/oarriaga/face_classification
• https://github.com/mikel-brostrom/yolo_tracking
• https://github.com/ultralytics/ultralytics
• https://github.com/derronqi/yolov8-face
• https://github.com/ReID-Team/ReID_extra_testdata
• https://data.mendeley.com/datasets/hvnsh7rwz7/1