Comparison between (a) traditional unsupervised method and (b) our proposed unsupervised method.

<p>We introduce a Transformer-based image pose encoder along with an additional similarity loss to enhance landmark detection capabilities.</p>

Saved in:

Bibliographic Details
Main Author:	Can Zhao (706105) (author)
Other Authors:	Tao Wu (16763) (author), Jianlin Zhang (440945) (author), Zhiyong Xu (3987659) (author), Meihui Li (1568122) (author), Dongxu Liu (547486) (author)
Published:	2024
Subjects:	Cell Biology Neuroscience Space Science Biological Sciences not elsewhere classified Information Systems not elsewhere classified siamese contrastive regularizer landmark detection tasks existing unsupervised methods siamese comparative transformer achieving semantic representations image pose encoder trained encoder semantic relationships guided transformer source image image reconstruction thereby improving lightweight direction common task based network
Tags:	Add Tag No Tags, Be the first to tag this record!

Similar Items