Clinical Tooth Color Recognition System Based on Deep Learning

Yen Ju Chen, Pi Ting Hsu, Cheng Yang Liu

Research output: Chapter in Book/Report/Conference proceedingConference contributionpeer-review

Abstract

The aim of this study is to develop a low-cost measurement system for tooth color identification. In this research, tooth identification is carried out using machine learning for locating the position of teeth and XGBoost for the color recognition of teeth. In the hardware, the capturing mechanism is designed using computer-aided design software that ensures the measurements taken from the same position. The dataset used for training in this study is captured by us, including 450 images of teeth models from the Vita classical A-series tooth shade guide. These images are divided into the 80% training set and the 20% validation set. The final results show that the object detection achieved an average precision of 0.995 (mAP50), while the tooth color recognition achieved an accuracy of 0.775. The overall measurement speed is approximately 8 milliseconds. This precise tooth color identification and rapid detection significantly reduce the time requirement for tooth model production.

Original languageEnglish
Title of host publication2023 International Conference on Next Generation Electronics, NEleX 2023
PublisherInstitute of Electrical and Electronics Engineers Inc.
ISBN (Electronic)9798350319088
DOIs
StatePublished - 2023
Event2023 IEEE International Conference on Next Generation Electronics, NEleX 2023 - Vellore, India
Duration: 14 Dec 202316 Dec 2023

Publication series

Name2023 International Conference on Next Generation Electronics, NEleX 2023

Conference

Conference2023 IEEE International Conference on Next Generation Electronics, NEleX 2023
Country/TerritoryIndia
CityVellore
Period14/12/2316/12/23

Keywords

  • artificial intelligence
  • deep learning
  • object detection
  • tooth shade

Fingerprint

Dive into the research topics of 'Clinical Tooth Color Recognition System Based on Deep Learning'. Together they form a unique fingerprint.

Cite this