AI REFEREE ASSISTANT
Virtual Assistant Referee is an AI-powered solution designed for football video analysis. It utilizes deep learning models like CNN to detect player contact and fouls in football videos in real-time.
Our project aims to enhance sports officiating by providing automated video analysis capabilities. With Virtual Assistant Referee, referees and coaches can quickly identify critical moments in games, improving decision-making and performance evaluation.
Procedure
1
Data Collection and Preparation
Gathered a diverse dataset of sports videos covering various games and scenarios, including player contact and fouls. Annotated the dataset to label instances of player contact and fouls, ensuring accurate training data for the models.
2
Model Development and Training
Developed deep learning models using frameworks like TensorFlow and Keras to detect player contact and fouls in sports videos. Trained the models on the annotated dataset, optimizing performance metrics such as accuracy, precision, and recall.
3
Integration and Deployment
Integrated the trained models into a user-friendly application interface using tools like Streamlit.
Deployed the application to enable real-time analysis of sports videos, providing instant feedback to referees and coaches during games and training sessions.
ABOUT THE PROJECT
How Does It Work?
Our VAR system leverages
two types of deep learning models:
V1 (RGB Rendering): Analyzes videos in full color, focusing on subtle details in player movement.
V2 (Grayscale Rendering): Examines player shapes and contact points, ideal for low-light or unclear footage.
Tech: Built using cutting-edge AI frameworks like TensorFlow and deployed with Streamlit for an interactive web experience.
TEAM
Team Lead
Joel Joseph
Role: Backend Developer
Responsibilities: Conducted model training, testing, and deployment.
Ensured the robustness and reliability of the project's technical infrastructure.
Team Member
A Ardra Nair
Role: Data Specialist
Responsibilities: Led data preprocessing and collection efforts. Ensured the availability of high-quality data for model training. Also contributed to presentations and project reports.
Team Member
Aswin E V
Role: Frontend and Backend Integration Specialist
Responsibilities: Skillfully integrated frontend and backend components, utilizing HTML and CSS to create an engaging user interface. Ensured seamless integration for an enhanced user experience.
Team Member
Aatlee V Varghese
Role: UI/UX Designer
Responsibilities: Led Streamlit design and other frontend activities, contributing to an attractive and user-friendly interface. Implemented innovative designs to enhance usability and attractiveness.
