DAEMLABS
Computer vision system that tracks & predicts basketball trajectory and detects hoop collisions using polynomial regression. Monitors R² drop below 0.99 to instantly classify rim hits versus successful shots.
Coaches and analysts lack an automated system to determine whether a basketball shot successfully enters the hoop or collides with the rim—forcing manual frame-by-frame video review that is time-consuming, subjective, and impossible to scale across multiple game footage clips.
We built a computer vision pipeline using OpenCV for real-time ball tracking across video frames, then applied 2nd-degree polynomial regression to model the parabolic projectile motion. By monitoring the R² coefficient of determination frame-by-frame—where a sudden drop below 0.99 indicates trajectory disruption—the system detects rim collisions versus successful hoop entries without manual intervention.
Real-time ball tracking using OpenCV color filtering & contour detection
R² threshold (0.99) effectively distinguishes rim hits from clean shots
2nd-degree polynomial regression accurately models & predicts parabolic basketball trajectories
Supports both game footage & simulated shot videos
Trajectory visualization outputs for debugging and post-game analysis
Zero manual frame labeling – fully automated collision detection