Human Identification with Dental Panoramic Images Based on Deep Learning

摘要

Human identification by means of dental panoramic X-ray images has been achieveddue to end-to-end deep learning using a convolutional neural network. This paperproposes a novel attention-based multi-supervision network (AMNet) for humanidentification. AMNet includes an attention-based supervision mechanism to obtainan accurate attention mask, feature fusion branches to combine multilevel globalfeatures, and part feature branches to obtain discriminative local features. Afterextracting features of dental panoramic X-ray images, matching scores between thegallery and query features are calculated with cosine similarity to determine if thequery image and gallery image are from the same identity. Our training dataset has22,172 images from 9490 subjects. The proposed method achieved 88.72% rank-1accuracy and 95.79% rank-5 accuracy on the query set with 665 dental panoramicX-ray images from 503 different subjects.