Codes of this repository are for papper entitled with "Semi-Supervised Feature Learning for Improving Writer Identification", which had been published on Information Sciences.
-
Matlab, Matconvnet, Opencv, NVIDIA GPU
-
(Note that I have included my Matconvnet in this repo, so you do not need to download it again. I has changed some codes comparing with the original version. For example, one of the difference is in
/matlab/+dagnn/@DagNN/initParams.m
. If one layer has params, I will not initialize it again, especially for pretrained model.)You just need to uncomment and modify some lines in
compile.m
and run it in Matlab. Try it~ (The code does not support cudnn 6.0. You may just turn off the Enablecudnn or try cudnn5.1)If you fail in compilation, you may refer to http://www.vlfeat.org/matconvnet/install/
At first, you segment the document to lines with statistical line segmentation. You can refer to guideline (https://github.com/KiM55/DLS-CNN/).
-
Make a dir called
data
by typingmkdir ./data
. -
Download ResNet-50 model pretrained on Imagenet. Put it in the
data
dir. -
Add your original dataset path and extra dataset path into
prepare_data.m
andprepare_extra_data.m
and run it. Make sure the code outputs the right image path. -
Run
train_id_net_res_market_wlsr.m
(Single) orresnet52_2stream_wlsr_fc_identify_solely.m
(2-Stream) for training the proposed method.
Run test/feature_extraction.m
to extract the features of images in the gallery and query set. They will store in a .mat file. Then you can use it to do evaluation.
Run evaluation/evaluation.m
for evaluation.
- [1] 'Unlabeled Samples Generated by GAN Improve the Person Re-identification Baseline in vitro' by Zhedong Zheng et. al, paper, code
- [2] 'A Robust Off-line Writer Identification Method' by Shiming Chen et. al, paper, code
If this work is helpful for you, please cite my paper.
@article{Chen2019Semi,
title={Semi-Supervised Feature Learning for Improving Writer Identification},
author={Chen, Shiming and Wang, Yisong and Lin, Chin-Teng Lin and Ding, Weiping and Cao, Zehong},
journal={Information Sciences},
volume={482},
pages={156-170},
year={2019}
}
If you run into any problems with these codes, please submit a bug report on the Github site of the project. For another inquries please contact with me: gchenshiming@gmail.com or g_shmchen@163.com