Tennis Match Tracker

This project uses computer vision and deep learning to track tennis players and the ball within match footage. It combines YOLO-based models for ball detection and player tracking, along with a ResNet50-based model to detect court keypoints. The system processes video frames, tracks player and ball locations, and visualizes the results with bounding boxes and keypoints.

Features:

• Ball Tracking: Detect and track the ball using YOLOv5.
• Player Tracking: Track players using a YOLOv8 model.
• Court Keypoint Detection: Detect court lines and key points using a ResNet50-based model.
• Frame Annotation: Visualize player and ball locations with bounding boxes and court keypoints.
• Output: Annotated video saved for further analysis.

Requirements:

• Python 3.x
• PyTorch
• OpenCV
• torchvision
• other dependencies (see requirements.txt)

Installation:

1. Clone the repository:

   git clone https://github.com/yourusername/tennis-match-tracker.git
   

2. Install dependencies:

   pip install -r requirements.txt
   

3. Download the required pre-trained models:

   • YOLOv5 model
   • ResNet50 Court Model
   • YOLOv8 model for player tracking

Usage:

1. Place your input video in the data/ folder.
2. Place your models in models/ folder.
3. Run the script:

   python main.py
   

4. The output video will be saved as video.avi.

Notes:

• Ensure that the model paths are correctly set in the script.
• The input video should be in .mp4 format.