Car Dent Detection Using Watershed Segmentation and Deep Learning

Deep learning-based car dent detection is a new topic with enormous promise for a variety of industries, including insurance and resale businesses. Human examination is used in traditional approaches to find automotive dents, although this can be time-consuming and error-prone. Deep learning improvements have made it possible to use computer vision algorithms to automatically detect dents. The system has the potential to be employed by insurance companies, car dealerships, and car repair shops to identify car dents quickly and accurately. The proposed method consists of two stages: object detection and segmentation. In the first stage, the yolov5 model is trained on a dataset of car images to detect the location of dents in the image. In the second stage, watershed segmentation is applied to the detected dent regions to segment the dent area accurately. The proposed method is evaluated on a dataset of car images containing different types of dents.