Face attribute classification based on pytorch

In this repository, we implement the face attribute classification method based on pytorch according to Adaptively Weighted Multi-task Deep Network for Person Attribute Classific

The official caffe code is https://github.com/qiexing/adaptive_weighted_attribute

[Paper]

Results

methods	test mean accuracy
resnet50	90.479
resnet50+adaptive weight	90.845

Download trained model

baiduyun: https://pan.baidu.com/s/18HLplvoyhp1RntQXiR84rA password:4zk4

Test

1. python test.py --img_path {target_image_path}

Train

1. Download img_celeba_aligned dataset

2. python train.py --input_path {img_celeba_aligned_path}