The vehicle damage assessment system (VDAS) computerizes vehicle damage assessment and estimation of repair costs by employing deep learning techniques. Automated system examines high-resolution pictures to recognize the kind of destruction like dents, scratches, and structural problems that frequently occurs on various vehicle. By integrating with a fixed cost system VDAS provides valid repair cost estimates that are immensely dependent on the degree of destruction upon the vehicle to enable insurance companies, motor workshops and vehicle owners make quick informed decisions. VDAS also provides simple user interface where customer can enter images of damaged vehicle quickly as well as gain detailed evaluations and price forecasts. Mask-RCNN and YOLOv5 models are utilized for efficient car and bike damage detection task which facilitates the accurate damage detection thereafter performing assessment depending upon the severity of damage to predict the total cost of repair work. YOLOv5 model achieves higher accuracy of 71.9% with an overall confidence F1-score across all vehicle damage classes which is 0.39 at a confidence threshold of 0.477.
Automated Vehicle Damage Detection and Repair Cost Estimation Using Deep Learning
2024-07-10
728133 byte
Conference paper
Electronic Resource
English
Elsevier | 2024
|