Skip to main content
Fig. 1 | BMC Oral Health

Fig. 1

From: Detecting representative characteristics of different genders using intraoral photographs: a deep learning model with interpretation of gradient-weighted class activation mapping

Fig. 1

Performance of the R-net model in testing set when dealing with uncropped (raw) images and cropped images where perioral soft tissues were removed. Results showed that the R-net model exhibited higher recall (sensitivity) in male detection and higher precision (specificity) in female detection in contrast to performance in detection of the other gender. For cropped images and images from individuals under 15 years old, performance of the R-net model was decreased. * p < 0.05

Back to article page

BMC Oral Health

ISSN: 1472-6831

Contact us