Skip to content

Latest commit

 

History

History

vpl

Folders and files

NameName
Last commit message
Last commit date

parent directory

..

backbones

backbones

 
 

configs

configs

 
 

eval

eval

 
 

utils

utils

 
 

README.md

README.md

 
 

dataset.py

dataset.py

 
 

eval_ijbc.py

eval_ijbc.py

 
 

inference.py

inference.py

 
 

losses.py

losses.py

 
 

onnx_helper.py

onnx_helper.py

 
 

onnx_ijbc.py

onnx_ijbc.py

 
 

run.sh

run.sh

 
 

torch2onnx.py

torch2onnx.py

 
 

train.py

train.py

 
 

vpl.py

vpl.py

 
 

README.md

Variational Prototype Learning for Deep Face Recognition

This is the Pytorch implementation of our paper Variational Prototype Learning for Deep Face Recognition which is accepted by CVPR-2021.

How to run

Define a new configure file such as configs/example_ms1m.py, and start the training process by:

bash run.sh configs/example_ms1m.py

Results

Results on WebFace600K(subset of WebFace260M), loss is margin-based softmax.

Backbone Dataset VPL? Mask Children African Caucasian South Asian East Asian MR-All
R50 WebFace600K NO 78.949 74.772 89.231 94.114 92.308 73.765 90.591
R50 WebFace600K YES 78.884 75.739 89.424 94.220 92.609 74.365 90.942