This paper introduces vision-based intelligent inspection system to detect surface defects in CNC-machined parts by using a YOLOv7-based transfer learning model.The manual inspection systems used traditionally are laborious, subjective, and prone to mistakes, thus necessitating automated solutions.The suggested YTL-ISDD system takes advantage of the high-resolution images taken under controlled lighting to detect defects, including scratches, cracks, pits, and burrs.The model makes use of pre-trained YOLOv7 weights that improve feature extraction and minimise the training time.The amount of data augmentation, such as rotation, scaling, flipping, and contrast adjustment, is used to enhance robustness and generalisation in different conditions of surfaces.The system can be easily integrated into CNC production environments, and it can detect defects in real-time, accurately, and consistently with minimum human involvement, which enhances the quality of products and efficiency of production.
Zhou et al. (Thu,) studied this question.