Multi-Object Tracking

This repository provides a end-to-end solution for Multi-Object Tracking.

Requirements

To set up the environment, clone the repository and install the necessary packages:

conda create --name obj_track_conda

conda activate obj_track_conda

pip3 install ipykernel

python3 -m ipykernel install --user --name obj_track_conda --display-name obj_track_conda

pip3 install -r requirements.txt

Data

The demo.mp4 from Internet (Only for demo purpose)

Please prepare your own video clip for demo.

Usage

To run the model on a video, use the following command:

python demo.py \
  --model_file      ./model/object_tracking.onnx \
  --video_path      ./data/veh2.mp4 \
  --line_points     6 793 1675 777 \
  --classes_interest person motor-vehicle non-motor-vehicle \
  --opts_dir        ./res \
  --conf_thres      0.1 \
  --iou_thres       0.7 \
  --draw_tracks \
  --draw_boxes

The scripts will save the visualized results.

Visualized Image

assets

Note: This is a demonstration. For higher accuracy, please customize the training strategy.

References

The torch weight is from Yolov8