F1Vision: Formula One Team Detection via Deep Learning

Objective:

Develop a robust AI system to automatically detect and classify Formula One cars into their respective teams, overcoming challenges such as strict aerodynamic standardization, high-speed motion blur, and dynamic lighting conditions.

Technical Approach:

Methodology: Transitioned from classical image processing to advanced Deep Learning-based Object Detection to capture subtle visual distinctions.

Infrastructure: Leveraged cloud-based GPU training for efficient model optimization and scalability.

Data Strategy: Implemented rigorous data augmentation to ensure model robustness against occlusions and varying race environments.

Key Results:

92.9% F1-Score across all 10 official teams.

Proven reliability in distinguishing highly similar objects (Fine-Grained Recognition) under real-world constraints.