Fine-Tuned YOLO Family Model Weights for Automatic Tooth Detection and FDI-Based Notation in Dental Panoramic Radiographs

Description

Model weights used in the study entitled "Performance Evaluation of YOLO Family for Tooth Detection and Segmentation on Panoramic Dental Radiographs".

Steps to reproduce

A total of 29,012 teeth were annotated with both bounding boxes and segmentation masks in 1,061 anonymized panoramic dental radiographs. The dataset was divided into training (70%), validation (20%), and test (10%) subsets. Five YOLO variants (for object detection YOLOv5x6u, YOLOv8x, and YOLOv11x, for instance segmentation YOLOv8x-seg and YOLOv11x-seg) were fine-tuned using input images resized to 1280×1280 pixels.

Categories

Licence