This study aims to eliminate the subjectivity and inconsistency inherent in the traditional International Association of Drilling Contractors (IADC) bit wear rating process, which heavily depends on the experience of drilling engineers and often leads to unreliable results. Leveraging advancements in computer vision and deep learning algorithms, this research proposes an automated detection and classification method for polycrystalline diamond compact (PDC) bit damage. YOLOv10 was employed to locate the PDC bit cutters, followed by two SqueezeNet models to perform wear rating and wear type classifications. A comprehensive dataset was created based on the IADC dull bit evaluation standards. Additionally, this study discusses the necessity of data augmentation and finds that certain methods, such as cropping, splicing, and mixing, may reduce the accuracy of cutter detection. The experimental results demonstrate that the proposed method significantly enhances the accuracy of bit damage detection and classification while also providing substantial improvements in processing speed and computational efficiency, offering a valuable tool for optimizing drilling operations and reducing costs.