Master Thesis Opportunity: Computer Vision for Motorsport Video Analysis

Your work will focus on transforming this raw video data into structured and actionable insights by improving current computer vision models.

Key tasks include:

• Reviewing and selecting state-of-the-art object detection and video analysis methods
• Designing and implementing improvements to an existing baseline model
• Enhancing robustness against real-world challenges such as: Occlusions, Lighting variability, Complex camera perspectives, Visually similar vehicles
• Evaluating performance improvements using objective metrics
• Generating structured outputs (e.g. vehicle detection and relative positioning) for further analysis
• Documenting findings, limitations, and recommendations for future development

Technologies & Environment

You will work in a modern machine learning environment, including:

• Python-based ML stack (e.g. PyTorch, OpenCV)
• Object detection frameworks (e.g. YOLO or similar)
• AWS-based data access and processing
• Large-scale motorsport video datasets from past WEC events

Do you have experience in Python?, Do you have a Master's degree?, * Master's student in Computer Science, Engineering, Data Science, Mathematics, Physics, or a related field

• Hands-on experience with machine learning and computer vision in Python
• Experience with at least one deep learning framework (preferably PyTorch)
• Basic understanding of object detection, image processing, or video analysis
• Experience with OpenCV, YOLO-style models, tracking, segmentation, or data annotation is a plus
• Strong analytical mindset and ability to evaluate models using objective metrics
• Interest in motorsport and applied AI systems is beneficial

At Toyota Racing, we are continuously advancing the use of AI and computer vision to unlock insights from complex race environments. We offer an exciting Master's thesis opportunity focused on state-of-the-art computer vision methods for analysing race video data in the FIA World Endurance Championship (WEC).The goal of this thesis is to enhance an existing automated race-video analysis system by improving its robustness, accuracy, and real-world performance using modern deep learning approaches.