⚽️ FOOTCVISION : Computer Vision Applied to Football

Roadmap 🛣️

Phase 1: Player Detection and Conformal Prediction 🏃‍♂️⚽️

1. YOLOv11 Fine-Tuning for Player Detection 🎯

2. Conformal Object Detection with puncc library 📏

---

Phase 2: Two Approaches for Team Differentiation 📊

• HSV Classifier:
  • Extracted HSV colors from bounding box regions of detected players.
  • Cluster players into teams based on dominant uniform colors.

• K-Means Clustering for Team Analysis:
  • Used player positions (bounding box coordinates) and CLIP features to cluster players into two teams.

---

Phase 3: Ball Tracking and Player Statistics 🎥⚽

1. Ball & Player Tracking ⚽ The first video is ball & player tracking with HSV Classifier and the second is from the Kmeans classifier.

2. Player Statistics 📈 Implemented the Metrics Class allowing ball possesion, computed the percentage of possesion of each team and marked the player being in possesion.

---

Phase 4: Offside Detection 🚩

1. Rule-Based Offside Detection

---

Installation and Usage 🚀

Clone the Project

git clone https://github.com/alyasltd/FootCVision.git

Install Dependencies

pip install -r requirements.txt