CV ProjectReal-Time Ball Tracking
Computer Vision Pipeline

Real-Time Ball Tracking and Trajectory Prediction

End-to-end CV pipeline for real-time ball detection, tracking, smoothing, trajectory prediction, and analytics export using YOLOv8 and Kalman Filtering.

01
Detection

Real-Time Ball Detection

Detects balls in video frames using YOLOv8 with high accuracy.

FPS60
02
Tracking

Multi-Ball Tracking

Assigns stable IDs and re-identifies balls across frames.

ID switchesMinimal
03
Smoothing

Kalman Filter Smoothing

Reduces jitter and provides smooth trajectory estimation.

Noise↓
04
Prediction

Trajectory Forecasting

Predicts future ball positions using physics-informed models.

Horizon1s
05
Export

Analytics Output

Exports tracked data in JSON/CSV for downstream analysis.

Formats2
Mean Absolute Error0.12
Throughput120 FPS
Latency<15 ms

Project Media

Tracking visualization

Real-Time Tracking Visualization

Trajectory visualization

Trajectory Heatmap