Colab code for image inpainting. The repository is based out of DeepFillv2 Pytorch awesome implementation. Link to the Original Paper
@article{yu2018generative,
title={Generative Image Inpainting with Contextual Attention},
author={Yu, Jiahui and Lin, Zhe and Yang, Jimei and Shen, Xiaohui and Lu, Xin and Huang, Thomas S},
journal={arXiv preprint arXiv:1801.07892},
year={2018}
}
@article{yu2018free,
title={Free-Form Image Inpainting with Gated Convolution},
author={Yu, Jiahui and Lin, Zhe and Yang, Jimei and Shen, Xiaohui and Lu, Xin and Huang, Thomas S},
journal={arXiv preprint arXiv:1806.03589},
year={2018}
}
Make sure to change runtime in colab to GPU for quick results!
NOTE
- The inpainting is being done after resizing the image to 512x512. This can be changed in the RESIZE_TO parameter in the config.py file.
- An example of an image and corresponding mask is present in the example folder of the repo. The notebook also allows generation of random masks. The uploaded image and the masks will be present in the input folder.
- This repo has not been tested for training. Inference works fine.
Check it out and let me know if any issues!