dentex-tooth-segmentation

1. Model Summary

• Purpose: Dental_008 FDI 32-Tooth Instance Segmentation Model (YOLOv8m-seg)
• Source Code Repository: https://github.com/HyunchanAn (MIT License)

2. Training Data & License Conditions

• Upstream Datasets: DENTEX Challenge Dataset (Hamamci et al., 2023, CC BY-NC-SA 4.0) leak-free 507/127 split (Box mAP@0.5 0.9915, Mask mAP@0.5 0.9855) + 60 edentulous hard negatives.
• Usage Restriction: Models trained on CC BY-NC-SA 4.0 (DENTEX) or restricted clinical/competition datasets are distributed strictly for non-commercial academic research and educational evaluation (cc-by-nc-sa-4.0). They must not be used for direct clinical diagnosis on patients without regulatory clearance or redistributed commercially.