Skip to content

nissansz/EraseNet

 
 

Folders and files

NameName
Last commit message
Last commit date

Latest commit

 

History

15 Commits
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 

Repository files navigation

EraseNet

This repository is the implementation of EraseNet, a neural network for end-to-end scene text removal.

Data preparation

The data preparation can be refer to ./examples/. You can download our datatset at SCUT-EnsText or synthetic dataset SCUT-Syn for training and testing.

SCUT-EnsText needs decompression password, you can send me at [email protected] for it.

Environment

Anaconda is recommended to establish a virtual environment to run our code. My environment can be refered as follows:

python = 3.7
pytorch = 1.3.1
torchvision = 0.4.2

Training

Once the data is well prepared, you can begin training:

python train_STE.py --batchSize 4 \
  --dataRoot 'your path' \
  --modelsSavePath 'your path' \
  --logPath 'your path'  \

Testing and evaluation

If you want to predict the results, run:

python test_image_STE.py --dataRoot 'your path'  \
            --batchSize 1 \
            --pretrain 'your path' \
            --savePath 'your path'

To evaluate the results:

python evaluatuion.py --target_path 'results_path' --gt_path 'labels_path'

Acknowledge

The repository is benefit a lot from LBAM and GatedConv. Thanks a lot for their excellent work.

Citation

If you find our method or dataset useful for your reserach, please cite:

@ARTICLE{Erase2020Liu,
  author     ={Liu, Chongyu and Liu, Yuliang and Jin, lianwen and Zhang, Shuaitao and Luo, Canjie and Wang, Yongpan},
  journal    ={IEEE Transactions on Image Processing},
  title      ={EraseNet: End-to-End Text Removal in the Wild},
  year       ={2020},
  volume     ={29},
  pages      ={8760-8775},}

@article{zhang2019EnsNet,
    title     = {EnsNet: Ensconce Text in the Wild},
    author    = {Shuaitao Zhang∗, Yuliang Liu∗, Lianwen Jin†, Yaoxiong Huang, Songxuan Lai
    joural    = {AAAI}
    year      = {2019}
  }

Feedback

Suggestions and opinions of our work (both positive and negative) are greatly welcome. Please contact the authors by sending email to Chongyu Liu([email protected]). For commercial usage, please contact Prof. Lianwen Jin via ([email protected]).

About

scene text removal

Resources

License

Stars

Watchers

Forks

Releases

No releases published

Packages

No packages published

Languages

  • Python 100.0%