rCounting the Future: Deep Learning-Based Electric Vehicle Estimation from Video Surveillance

Abstract

Research and policymaking activity on sustainable transportation thrust the importance of electric vehicle (EV) existence and distribution within urban environments. A computer vision framework presented in this study enables video surveillance data to determine EV counts within particular geographical locations. The methodology uses YOLOv11 license plate recognition together with YOLOv5 object detection to explore electric vehicle plates during examination and vehicle identification processes. A precise vehicle and license plate cropping process maintains accuracy. Processed information is applied to the recognition of electric versus non-electric vehicles. The system proves efficiency by using video surveillance from public roads to identify vehicles, recognize their license plates, and count electric vehicles. An intricate two-stage system, “search” and “identify” phases, ensures proper vehicle and license plate cutting. Further processing of the data completes the identification of conventional and electric vehicles. The system shows its functionality by using public road video footage to identify cars and extract license plates before counting electric vehicles. The investigation demonstrates that deep learning enables both traffic monitoring through analysis and environmental assessment.