🚀 Exciting update! We have created a demo for our paper, showcasing the adaptive removal capabilities of our method. Check it out here!

DeepEraser

🔥 Good news! Our work has been accepted by IEEE Transactions on Multimedia (TMM), 2024.

The official code for “DeepEraser: Deep Iterative Context Mining for Generic Text Eraser”.

🚀 Demo (Link)

1. Upload the image to be erased in the left box.
2. Draw the mask over the text to be erased on the image.
3. Click the "Submit" button.
4. The output image will be displayed in the right box.

Inference

We have already released the pre-trained model for the SCUT-ENSTEXT dataset, i.e., $ROOT/deeperaser.pth. The pre-trained models for the three datasets in the paper are available at the Google Drive.

1. Put the distorted images in $ROOT/input_imgs/ and rename it to input.png.
2. Put the mask image in $ROOT/input_imgs/ and rename it to mask.png.
3. Run the script and the processed image is saved in $ROOT/output_imgs/ by default.

   python demo.py
   

Citation

If you find this code useful for your research, please use the following BibTeX entry.

@article{feng2024deeperaser,
  title={DeepEraser: Deep Iterative Context Mining for Generic Text Eraser},
  author={Feng, Hao and Wang, Wendi and Liu, Shaokai and Deng, Jiajun and Zhou, Wengang and Li, Houqiang},
  journal={IEEE Transactions on Multimedia},
  year={2024}
}