tyler3490 / tennisproject

Tennis Project

Intro

This project is the final project for B.Sc degree at the Technion institute. In this project I present an end-to-end model that uses computer vision algorithms and deep learning networks in order to process and analyse official tennis matches. Main paper used for this project is Vinyes-Mora-S-2018-PhD-Thesis

Project Scope

This project is mainly focused on official match videos, and camera position and angle as shown in the next section.

Example for some tasks completed in this project are:

1. Detecting and tracking the tennis court in frame
2. Detecting and tracking both player in the frame
3. Detecting and tracking the ball
4. Extract the bottom player skeleton
5. Detecting the exact moment of strokes
6. Classify the strokes of the bottom player
7. Create top view gameplay
8. Calculate statistics of the gameplay of both players including positions heatmap and distance travelled

Example

Requirements

Python 3.4, OpenCV (cv2), PyTorch and other common packages listed in requirements.txt at the src directory.

Entry point

Entry point for the code is in process.py (scroll to the bottom to see main function video_process(...)).