Adaptive traffic control systems (ATCSs) adjust traffic signal timing based on demand. We begin by presenting the main building blocks of a vision-based ATCS, including pre-processing, vehicle detection, vehicle classification and vehicle tracking. Next, we examine several of the key technical challenges in developing a computer vision-based ATCS and explore approaches for overcoming these challenges. These challenges stem from the need for an ATCS to perform accurate person and vehicle detection despite a huge variety of vehicle types, occlusion of objects of interest and difficult lighting conditions.
Challenges in Vision-Based Adaptive Traffic Control Systems
Session Speakers
-
Venkatesh Wadawadagi
Solution Consultant, Sahaj Software Solutions
Venkatesh Wadawadagi is a Principal Data Scientist, Practice Leader, Speaker, Author and Trainer with 9+ years of hands-on domain and technology experience in R&D and product development; specializing in Visual-AI, Embedded-AI, Software Engineering and Multimedia Subsystems. As a solution consultant at Sahaj Software Solutions, he helps businesses solve complex problems using AI-powered solutions. He has extensive experience in research, building algorithms, leading the teams/efforts and architecting the solution. He has worked on a variety of solutions ranging from building brand monitoring system, building a Deep Learning Inference Engine for an e-AI chip, implementing Worker Safety Monitoring System to research and development in the area of ATM surveillance.
