Automated Damage Detection and Repair Cost Estimation for Automobiles

Estimation of vehicle damage and repair cost is extremely important in the current automobile and insurance industries. Conventional inspection techniques are time-consuming, subjective, and susceptible to inconsistency. In this paper, a method for an autonomous surface-level vehicle damage estimation and repair cost estimation using a deep learning technique called Mask R-CNN is proposed. The system computes input images for damage localizing and feature extracting like damage type, damage location, and damage size on which cost calculation is performed. The proposed model attainted accuracy of 88.78% with an F1-score of 0.880 and thereby showcased efficiency in object segmentation as well as cost estimation. The system cuts down enormously the need for human interaction, improves consistency, and presents a scalable solution to apply on real-world applications of automobile repair and insurance claim adjustment.