Overview This repository contains a Python program utilizing YOLOv8 from Ultralytics to perform team prediction and player detection in a custom dataset. The program achieves an mAP of 38 at 50 IoU (Intersection over Union) and maintains an average FPS (Frames Per Second) range of 18-20.
- Team Prediction: Identifies and predicts teams based on specified criteria.
- Player Detection: Detects and localizes players within a given frame or set of frames.
- Custom Dataset: Trained and tested on a custom dataset for specific use cases.
- YOLOv8 Integration: Utilizes the YOLOv8 model architecture for efficient object detection.
- Performance Metrics: Achieves an mAP of 38 at 50 IoU on the custom dataset.
- FPS Monitoring: Sustains an average FPS range of 18-20 for real-time or batch processing.
- Python 3.x
- PyTorch
- Ultralytics YOLOv8
- git clone https://github.com/Ali-Fartout/Soccer-Vision.git
- run capture.py