Measurement-Oriented Micro-Burr Inspection Using Structure-Aware Imaging and Segmentation

Counts differ because each database indexes a different set of publications. We treat OpenAlex as the canonical count; Google Scholar is not shown (no API, and crawling it violates its ToS).

Micro-burr inspection is an important quality-control task during dental bur manufacturing. However, vision-based methods specifically developed for dental bur micro-burr inspection have received limited attention, particularly for real-time industrial applications. To address this problem, this paper proposes a structure-aware micro-burr inspection framework integrating end-face region extraction, instance segmentation, and embedded deployment. According to the structural characteristics of dental burs, a structure-aware region extraction strategy is first adopted to focus the inspection on the shank end-face region, where burr defects are exclusively distributed. The extracted end-face images are subsequently processed by a YOLOv8-seg network to achieve pixel-level burr localization. The predicted segmentation masks are directly used to determine the presence or absence of burr defects for automated quality inspection. Experimental results demonstrate that the proposed method achieves a Precision of 0.989, a Recall of 0.905, and an mAP@0.5 of 0.969 on the test dataset. Furthermore, the optimized model is successfully deployed on an RK3588 embedded platform, achieving approximately 9 FPS at a 640 × 640 input resolution for real-time inspection. The proposed framework provides an effective vision-based solution for automated dental bur micro-burr inspection and demonstrates the feasibility of combining structure-aware imaging, instance segmentation, and embedded artificial intelligence for practical manufacturing applications.