The offical implementation of PERT: What is the Real Need for Scene Text Removal? Exploring the Background Integrity and Erasure Exhaustivity Properties (TIP2023). PERT progressively erases the text region with a balanced multi-stage erasure and Region-base Modification Strategy. The model is simple to be implemented (light weight) and can be easily developed.
10/8 The model is released
10/8 The code is released.
| Dataset | Model | PSNR | MSSIM | MSE | AGE | pEPs | pCEPs |
|---|---|---|---|---|---|---|---|
| SCUT-EnsText | Paper | 33.62 | 97.00 | 0.0013 | 2.1850 | 0.0135 | 0.0088 |
| SCUT-EnsText | This implementation | 34.12 | 97.06 | 0.0012 | 2.1299 | 0.0125 | 0.0080 |
| Dataset | Model | BI | EE |
|---|---|---|---|
| SCUT-EnsText | Paper | 63.55 | 80.00 |
| SCUT-EnsText | This implementation | 64.73 | 79.46 |
python=3.6.9 & torch==1.8.1+cu111
It is the best to use our provided Dockerfile for quick start.
Download SCUT-EnsText and put it into your folder.
Download our trained model from Google and put it into your folder.
bash test_Pert.shpython evaluatuion_bi.py --target_path /path_to_output --gt_path /path_to_label --BI Truepython evaluatuion_bi.py --target_path /path_to_output --gt_path /path_to_label --BI Falsebash train_Pert.shIf you find our method useful for your reserach, please cite
@ARTICLE{10214243,
author={Wang, Yuxin and Xie, Hongtao and Wang, Zixiao and Qu, Yadong and Zhang, Yongdong},
journal={IEEE Transactions on Image Processing},
title={What is the Real Need for Scene Text Removal? Exploring the Background Integrity and Erasure Exhaustivity Properties},
year={2023},
volume={},
number={},
pages={1-1},
doi={10.1109/TIP.2023.3290517}}