A Vision-Driven Framework for Evaluating Vehicle Damage Severity in Traffic Accidents
Year: 2025
Type: article
Abstract: Road traffic accidents remain a major global concern, particularly in developing countries where rising vehicle density and urbanization increase the risk of collisions. Traditional impact analysis methods-based on manual reporting are often slow, error-prone, and inconsistent. This study presents an AI-based system that utilizes image processing and Convolutional Neural Networks (CNNs) to automatically classify accident severity (minor, moderate, severe) from visual inputs such as CCTV and dashcam footage. The methodology includes image acquisition, preprocessing, model design, and evaluation using metrics like accuracy, precision, recall, and F1-score. Results demonstrate the system’s high performance and potential to enhance the speed, accuracy, and scalability of road accident analysis.
Author: Tanveer Alam
Language: English
Subfield: Computer Vision and Pattern Recognition
Field: Computer Science
Domain: Physical Sciences
Open Access status: green
OpenAlex ID: W6893315474
FWCI:
Cites:
Cited by:
Related to: