Paper Title
Teeth Detection and Caries Segmentation in Dental Panoramic X-Ray Using Yolov7

Abstract
This study primarily applies the YOLOv7 model for tooth detection and caries segmentation on panoramic dental images to effectively improve the accuracy and reliability of caries identification. A total of 400 publicly available dental panoramic images were used with a total of 3,399 teeth, of which 595 showed caries. The carious regions were manually annotated by two experienced dental specialists. To enhance the tooth contours, bilateral filtering was applied to the images prior to model training. Two separate models were trained using YOLOv7 for tooth detection and caries segmentation, respectively. The models’ performance was evaluated based on precision, recall, and mean average precision (mAP).The dental detection model had a model precision of 0.96, a recall of 0.98, and an average precision (mAP@0.5) of 0.95, while the caries segmentation model achieved precision of 0.78, a recall of 0.96, and an average accuracy (mAP@0.5) of 0.70. The results demonstrate that YOLOv7 exhibits strong performance in both tooth detection and caries segmentation tasks. The proposed system can automatically identify and accurately segment carious lesions in dental images, thereby enhancing clinical diagnostic efficiency for dentist. Keywords: Panoramic X-ray, Dental caries, Machine learning, Deeplearning, Object detection, Segmentation, Artificial intelligence, YOLO, Convolutional neural networks