# Intrinsic Decomposition of Document Images In-the-Wild

arXiv: https://arxiv.org/abs/2011.14447 · HTML: https://mlsanity.com/paper/2011.14447/

## Key facts

- Cited by 12 papers in the map (data through 14 Oct 2026).
- First posted on arXiv 29 Nov 2020.

## Built on by (most cited first, 12 of 12)

- [DocRes: A Generalist Model Toward Unifying Document Image Restoration Tasks](https://mlsanity.com/paper/2405.04408/) (2024)
- [DocScanner: Robust Document Image Rectification with Progressive Learning](https://mlsanity.com/paper/2110.14968/) (2021)
- [Deep Unrestricted Document Image Rectification](https://mlsanity.com/paper/2304.08796/) (2023)
- [LP-IOANet: Efficient High Resolution Document Shadow Removal](https://mlsanity.com/paper/2303.12862/) (2023)
- [Uni-DocDiff: A Unified Document Restoration Model Based on Diffusion](https://mlsanity.com/paper/2508.04055/) (2025)
- [DocAligner: Annotating Real-world Photographic Document Images by Simply Taking Pictures](https://mlsanity.com/paper/2306.05749/) (2023)
- [LayeredDoc: Domain Adaptive Document Restoration with a Layer Separation Approach](https://mlsanity.com/paper/2406.08610/) (2024)
- [GL-PGENet: A Parameterized Generation Framework for Robust Document Image Enhancement](https://mlsanity.com/paper/2505.22021/) (2025)
- [SyntheticDoc: A Large Synthetic Dataset for Document Unwarping and Illumination Correction](https://mlsanity.com/paper/2609.15503/) (2026)
- [ENet-GP: Unified Document Image Restoration](https://mlsanity.com/paper/2609.33758/) (2026)
- [After a Decade: Bringing Shadow Removal into the Real World with Agentic Training Data](https://mlsanity.com/paper/2609.38607/) (2026)
- [Rethinking Supervision in Document Unwarping: A Self-consistent Flow-free Approach](https://mlsanity.com/paper/2610.15445/) (2026)

## Abstract (arXiv)

Automatic document content processing is affected by artifacts caused by the shape of the paper, non-uniform and diverse color of lighting conditions. Fully-supervised methods on real data are impossible due to the large amount of data needed. Hence, the current state of the art deep learning models are trained on fully or partially synthetic images. However, document shadow or shading removal results still suffer because: (a) prior methods rely on uniformity of local color statistics, which limit their application on real-scenarios with complex document shapes and textures and; (b) synthetic or hybrid datasets with non-realistic, simulated lighting conditions are used to train the models. In this paper we tackle these problems with our two main contributions. First, a physically constrained learning-based method that directly estimates document reflectance based on intrinsic image formation which generalizes to challenging illumination conditions. Second, a new dataset that clearly improves previous synthetic ones, by adding a large range of realistic shading and diverse multi-illuminant conditions, uniquely customized to deal with documents in-the-wild. The proposed architecture works in a self-supervised manner where only the synthetic texture is used as a weak training signal (obviating the need for very costly ground truth with disentangled versions of shading and reflectance). The proposed approach leads to a significant generalization of document reflectance estimation in real scenes with challenging illumination. We extensively evaluate on the real benchmark datasets available for intrinsic image decomposition and document shadow removal tasks. Our reflectance estimation scheme, when used as a pre-processing step of an OCR pipeline, shows a 26% improvement of character error rate (CER), thus, proving the practical applicability.