tf_angular_loss
tf_angular_loss copied to clipboard
Tensorflow Implementation of Deep Metric Learning with Angular Loss
Angular loss for deep metric learning
Tensorflow Implemenation of Deep Metric Learning with Angular Loss. https://arxiv.org/abs/1708.01682
NOTE:
- All classes in an input batch should be different.
- All embeddings should not be l2 normalized.
- When you guys use this angular loss with N-pair loss, let
with_l2_reg
beFalse
. If not, let it beTrue
. - I implemented only an angular mc loss. triplet version will be updated.
Reference
@inproceedings{wang2017deep,
title={Deep metric learning with angular loss},
author={Wang, Jian and Zhou, Feng and Wen, Shilei and Liu, Xiao and Lin, Yuanqing},
booktitle={2017 IEEE International Conference on Computer Vision (ICCV)},
pages={2612--2620},
year={2017},
organization={IEEE}
}