Details about Vehicle Specific Detection

This project focused on training the YOLOv8 model on the BDD100k dataset for vehicle-specific object detection, achieving a mean Average Precision (mAP) of 40.4%. In addition to detection, the project integrated ByteTrack and BoT SORT algorithms to track objects over time, with ByteTrack outperforming in maintaining object identities within the video streams. Despite facing limited CPU resources, the project successfully optimized the model’s performance, contributing to advancements in real-time vehicle detection and tracking for surveillance and autonomous driving systems.