Papers

Paper Year Code Unwarp Light
DocUNet: Document Image Unwarping via A Stacked U-Net 2018 Pytorch
DewarpNet: Single-Image Document Unwarping With Stacked 3D and 2D Regression Networks 2019 Pytorch
DocProj: Document Rectification and Illumination Correction using a Patch-based CNN 2019 Pytorch
Rectinet: A Gated and Bifurcated Stacked U-Net Module for Document Image Dewarping 2020 Pytorch
RectiNet-v2: A stacked network architecture for document image dewarping 2021  
DocTr: Document Image Transformer for Geometric Unwarping and Illumination Correction ⭐⭐ Oct 2021 Coming soon

Unwarp:

  • Problem formulation:
    • Input: RGB image (img of 3 channels)
    • Output: Flow displacement on the x and y axis (img of 2 channels)
      • The pixel-level 2D displacement vector field between the distorted image and the corresponding corrected image.
      • Regression per pixel

Dataset

  • https://www.kaggle.com/c/denoising-dirty-documents/data
  • https://stackoverflow.com/questions/62492457/what-could-be-the-solution-for-automatically-document-image-unwarping-caused-by
  • https://github.com/thomasjhuang/deep-learning-for-document-dewarping
  • https://expressexpense.com/blog/free-receipt-images-ocr-machine-learning-dataset/
  • https://way2vat.com/receipt-data-augmentation-using-virtual-people-for-real-image-analysis/
  • http://kth.diva-portal.org/smash/get/diva2:1463587/FULLTEXT01.pdf