Road Object Detection

This project was developed for the Autonomous Vehicles for Bangladesh Roads Kaggle competition. The challenge provided an object detection dataset featuring diverse driving environments spanning nine districts of Bangladesh.

Our team leveraged and fine-tuned the YOLOv6 model to accurately detect and predict the positions of 13 distinct object classes commonly found on Bangladeshi roads. These included vehicles such as auto-rickshaws, bicycles, buses, cars, trucks, and trains, among others. By adapting YOLOv6 to this unique dataset, we tackled the complexities of varied road conditions, local traffic patterns, and diverse object appearances, aiming to advance the capabilities of autonomous vehicle systems in real-world Bangladeshi contexts.