Published Paper Details:

This project presents an integrated perception system for autonomous vehicles, leveraging the YOLOv8 deep learning model for real-time object detection, vehicle tracking, and traffic sign recognition. The system enhances situational awareness by accurately detecting dynamic road elements such as cars, buses, motorcycles, pedestrians, and static hazards like potholes and traffic signs. The image detection module effectively identifies multiple indoor objects, including persons, cups, chairs, potted plants, and laptops, achieving confidence scores above 0.5 and demonstrating low-latency processing suitable for static environments. In outdoor scenarios, the system successfully detects potholes with confidence scores ranging from 0.28 to 0.83, clearly marking even small or partially obscured hazards, thus improving road hazard awareness for autonomous navigation. The vehicle tracking module assigns unique IDs to moving vehicles, such as cars and buses, and continuously monitors their trajectories across video frames. This allows the system to predict movement directions and support collision avoidance in dynamic traffic conditions. Additionally, the system reliably identifies various traffic signs with high confidence, including gap in median, speed limits, give way, and width restriction signs. This enables autonomous vehicles to adjust navigation and speed, respond to road priorities, and plan paths through constrained spaces. Overall, the proposed YOLOv8-based framework demonstrates robust, real-time performance across diverse environments, significantly enhancing the safety and decision-making capabilities of autonomous driving systems