Enhancing Dental Caries Detection with YOLOv9: A Comprehensive Analysis and Validation of an Automated Object Detection Model

This study presents the development and validation of an advanced object detection model based on YOLOv9 for the automated identification of dental caries. Utilizing a dataset of 270 dental images sourced from Kaggle, this proj ect introduces a methodological framework that includes image preprocessing, augmentation, and annotation to address the challenge of limited data. The dataset was expanded through augmentation techniques such as flipping, rotation, shearing, and noise addition, resulting in a final dataset size that supports robust model training and evaluation. Employing the YOLOv9 model, known for its speed and accuracy, we conducted extensive training, optimization, and validation processes. Our results showcase a significant improvement in model performance, with a mean Average Precision (mAP) increase from 59.6% to 93.485%, alongside substantial gains in precision, recall, and F1 scores for detecting dental caries. This study not only highlights the effectiveness of YOLOv9 in dental anomaly detection but also sets a precedent for employing advanced object detection models in medical imaging analysis, offering potential enhancements in diagnostic accuracy and efficiency.