Semantic Understanding of Road Scenes

This project is exploring the methodology to improve the performance of the current semantic segmentation models when encountering unexpected situations on road. We hope to ensure the autonomous vehicles could make right decisions quickly when running into construction zones and other unknown road conditions.