AI Powered Screening & Self-Diagnosis in Oral Health

This research work describes a deep learning-based solution that uses the MobileNetV2 architecture to diagnose dental diseases such as dental caries, calculus and gingivitis using intraoral colour pictures. The model uses multi-class classification for all the classes both diseases class and a healthy class with high accuracy while using minimal computer resources, giving it a low-cost alternative to standard diagnostics. Furthermore, a segmentation module which is made using traditional U-Net architecture, is included to offer exact localization of damaged regions using bounding boxes, improving diagnostic precision and interpretability. The dataset used to support the model has been carefully chosen to encompass a wide range of demographic characteristics, ensuring robustness and applicability across populations. This complete method has enormous promise to improve dental healthcare by providing efficient, accurate, and easily accessible diagnostics for common dental problems.