The goal of this project is to use computer vision to track and record the trajectory of juggling balls, in order to provide some insights to the juggler about the inconsistency in his juggling pattern.
This project relies heavily on the python implementation of opencv.
We recommend using virtualenv to keep clean dependencies for each project. To install all dependencies in your environment, simply run:
pip install -r requirements.txt
To run the tracker on a video, execute the following command:
python track.py FILE_PATH [--start-frame INT] [--end-frame int] [--movement-path PATH]
[--keypoints-path PATH] [--trajectories-path PATH] [--save-fps INT] [--display-fps INT]
[--keypoints-method METHOD]
For a complete detail of the output of each step of the algorithm, we recommand specifying
--movement-path --keypoints-path and --trajectories-path to get the complete output.
For more information on how to use the arguments, run python track.py --help
The data folder contains four test videos:
- juggling_front.mp4
- juggling_side.mp4
- juggling_higher.mp4
- juggling_low_quality.mp4
These four videos allow to test the tracker on different juggling angles, heights, and video qualities.
Test results can be found in the output folder.