JumpVision is a machine vision-based project for estimating an individual's jump height and hang time from video footage. By analyzing videos captured from a fixed camera angle, the system can track key body points and use physics-based calculations to determine jump performance.
- 📹 Video-based measurement: Uses computer vision to analyze jump motion.
- 🎯 Accurate jump height estimation: Calculates vertical leap based on body keypoint tracking.
- ⏱ Hang time calculation: Determines the total air time of the jump.
- 📊 Data visualization: Plots motion curves and key parameters for better analysis.
git clone https://github.com/SportsCV/JumpVision.git
cd JumpVision
pip install -r requirements.txt- Place a video file of a jump attempt in the project directory as
demo.mp4. - Run the analysis script:
python main.py- View the output jump height and hang time.
78.71cm
4. Compared with MY-JUMP-LAB APP
76.34cm
- Keypoint Detection: Uses a pose estimation model to track the ankle position.
- Jump Segmentation: Filters the frames to extract the jump phase.
- Parabolic Fitting: Applies a quadratic function to estimate jump height.
- Physics-based Calculation: Computes the total hang time and vertical leap.