Car body damage detection and cost estimation based on generative AI

Abstract

This research presents a robust AI-driven system for car body damage detection, impact assessment, and cost estimation, addressing critical challenges in the automotive repair and insurance claims process for car body panel damages. Leveraging Generative Artificial Intelligence (GAI) like GPT-4o, the study aims to automate damage identification and repair cost estimation. A comprehensive prompt exceeding 1,200 words was developed, incorporating cost estimation rules, predefined panel and damage classes, example outputs, and step-by-step instructions to guide the model. The system effectively identifies vehicle types, body panels, and damage types while accurately estimating damage severity and repair costs. To address limitations in cost estimation, the research introduces a rule-based framework aligned with industry standards for damage visual inspection, integrating expert insights and real-world datasets. Results demonstrate high accuracy in damage detection and cost estimation, significantly reducing processing times compared to traditional methods. Validated against expert assessments, the proposed system offers a scalable, efficient, and practical solution for the automotive and insurance sectors.