Research and Design of Real-Time Detection System for Car Surface Damage Based on YOLOv7 and Edge Devices

This paper presents a research and design of a real-time detection system for car surface damage based on YOLOv7 and edge devices. The system utilizes the high-performance computing capability and low power consumption of embedded GPUs to achieve detection and classification of vehicle surface damage on edge devices. This improves the efficiency and accuracy of damage assessment, meeting the requirements of digital transformation. The paper includes the following aspects: Firstly, the network architecture and loss function of YOLOv7 model are introduced, analyzing its advantages and applicability in vehicle surface damage detection. Secondly, the characteristics and selection of embedded GPU devices, as well as common types of vehicle surface damage, are discussed. Finally, the paper designs an edge device architecture based on embedded GPUs, including hardware and software aspects, and describes the steps of image preprocessing, feature extraction, object detection, damage classification, and result output.