This repository implements the ASTER in pytorch. Origin software could be found in here.

ASTER is an accurate scene text recognizer with flexible rectification mechanism. The research paper can be found here.

conda env create -f environment.yml

[NOTE] Some users say that they can't reproduce the reported performance with minor modification, like 1 and 2. I haven't try other settings, so I can't guarantee the same performance with different settings. The users should just run the following script without any modification to reproduce the results.

bash scripts/stn_att_rec.sh

bash scripts/main_test_all.sh

The pretrained model is available on our release page. Download demo.pth.tar and put it to somewhere. Before running, modify the --resume to the location of this file.

At present, the bidirectional attention decoder proposed in ASTER is not included in my implementation.

You can use the codes to bootstrap for your next text recognition research project.

We give an example to construct your own datasets. Details please refer to tools/create_svtp_lmdb.py.

If you find this project helpful for your research, please cite the following papers:

@article{bshi2018aster,
  author  = {Baoguang Shi and
               Mingkun Yang and
               Xinggang Wang and
               Pengyuan Lyu and
               Cong Yao and
               Xiang Bai},
  title   = {ASTER: An Attentional Scene Text Recognizer with Flexible Rectification},
  journal = {IEEE Transactions on Pattern Analysis and Machine Intelligence}, 
  volume  = {}, 
  number  = {}, 
  pages   = {1-1},
  year    = {2018}, 
}

@inproceedings{ShiWLYB16,
  author    = {Baoguang Shi and
               Xinggang Wang and
               Pengyuan Lyu and
               Cong Yao and
               Xiang Bai},
  title     = {Robust Scene Text Recognition with Automatic Rectification},
  booktitle = {2016 {IEEE} Conference on Computer Vision and Pattern Recognition,
               {CVPR} 2016, Las Vegas, NV, USA, June 27-30, 2016},
  pages     = {4168--4176},
  year      = {2016}
}

IMPORTANT NOTICE: Although this software is licensed under MIT, our intention is to make it free for academic research purposes. If you are going to use it in a product, we suggest you contact us regarding possible patent issues.