Prompting Vision-Language Models for Dental Notation Aware Abnormality Detection

Chenlin Du^1,*, Xiaoxuan Chen^2,*, Jingyi Wang², Junjie Wang³, Zhongsen Li¹, Zongjiu Zhang^1,📧, Qicheng Lao^3,📧.

¹THU    ²West China Hospital of Stomatology    ³BUPT

🏷️ TODO

• Release annotations.
• Release inference code.
• Release checkpoints.
• Release trainning code.

Installation

As we utilized the MMDetection implementation of GroundingDINO as our base model. Please refer to Installation and the Config File for installation instructions.

Data

Our work utlized the quadrant-enumerated subset of the DENTEX dataset, which we further annotated with six different dental abnoramlities. Please refer to their repository for the panoramic X-ray images download.

We thank the authors of DENTEX dataset for providing the valuable data.

♥️ Acknowledgement

Our model is related to GroundingDINO and MMDetction. Thanks for their great work!

We also thank all the dentists who have helped to annotate the data.