Multi-Modal Vehicle Detection and Recognition Using Deep Learning Techniques for Autonomous Driving Applications

Abstract

A sophisticated campus infrastructure should also comprise of efficient parking and vehicle tracking. The use and application of real time vehicle detection and counting system within a campus of a college to track vehicle and measure the availability of parking space. The system employs modern computer vision models and uses an object detective model, a deep learning-based (YOLOv8), and successfully identifies and tracks automobiles in real-time video streams captured at the campus entrances.
The proposed approach is dynamic in that it compares the number of vehicles arriving at the parking site and parking available places continuously with the number of vehicles arriving constantly. In order to help the campus security and administrative personnel in managing the traffic, routing traffic and rerouting it where required, the system sends automated notifications upon reaching the maximum capacity allowable by the number of vehicles in the campus or at a given location. The implementation demonstrates the practical applicability of AI-powered solutions into campus management to make it more efficient, less inclined to manual labor, and scalable to department-level usage of larger institutions or in cities.