Soccer Player Detection Project

Introduction

This project focuses on developing a machine learning model for detecting soccer players, balls, and referees on the pitch using a finetuned Faster R-CNN. We experimented with different datasets and compared results of different models in order to see the progress of our work.

Dataset

Selection and Preparation

1. Dataset Source: We chose a dataset from Roboflow for football player detection. Link to dataset.

2. Preparation:

   • Unzipped the dataset directly on Google Colab.
   • Reformatted the dataset to separate annotations and images into different folders.
   • Split the dataset into train, validation, and test sets.

Custom Dataset Class

• Created a custom VOC dataset class to work with the Faster R-CNN model, ensuring the dataset was properly formatted for training.

Model Training and Evaluation

Initial Implementation

1. Model Choice: We selected Faster R-CNN based on extensive literature review and discussions with our mentor.
2. Initial Testing: Implemented and tested Faster R-CNN on built-in PyTorch datasets to understand its performance.

Data Loader

• Developed a data loader to handle the custom dataset format, enabling proper loading and augmentation of training data.

Training

1. Initial Training: Trained the model on a small subset of images to verify the implementation.
2. Metrics Calculation:
   • Implemented code to calculate confusion matrix elements (True Positive, False Positive, False Negative) for each class.
   • Evaluated the initial model’s performance using these metrics.

Fine-Tuning and Performance

1. Fine-Tuning: Trained the model on the complete dataset using Paperspace, saving intermediate models after each epoch.
2. Results: Achieved a high detection accuracy (90-100%) for players, goalkeepers, and referees. The model also successfully detected balls when visible.
3. Issues: Identified issues with false positives, especially misclassifying goalkeepers and referees as players.

Visualization and Demo

1. Confusion Matrix: Plotted confusion matrices to visualize model performance.
2. Demo Creation: Created a demo by running a video through the trained model to visualize real-time performance. Identified several issues:
   • Misclassification of advertisements as players.
   • Overlapping players losing bounding boxes.
   • Smear frames causing detection failures.

Future Work

1. Improved Visualization: Develop a program to create graphical representations of class overlaps.
2. Team Detection: Plan to implement a method to detect different teams based on dominant colors within bounding boxes.
3. YOLO Exploration: Investigate YOLO models for potential real-time detection improvements, focusing on versions better suited for detecting small objects and handling multiple bounding boxes per grid cell.

Challenges with YOLO

1. Initial Attempts: Attempted to implement YOLOv1 but faced issues with small object detection and grid cell limitations.
2. Future Plans: Explore newer YOLO versions with pre-configured datasets to simplify the process and improve detection capabilities, most likey YOLOv8.

Conclusion

This project successfully developed a model for detecting soccer players, balls, and referees using Faster R-CNN. While initial results were promising, further improvements and exploration of alternative models like YOLO are planned to enhance performance and address current limitations.