# Lights, Camera, Action! A Framework to Improve NLP Accuracy over OCR documents

arXiv: https://arxiv.org/abs/2108.02899 · HTML: https://mlsanity.com/paper/2108.02899/

## Key facts

- Cited by 11 papers in the map (data through 14 Oct 2026).
- Builds on 9 of its 27 references that are in the map.
- First posted on arXiv 6 Aug 2021.

## Builds on (most cited first, 9 of 9)

- [Attention Is All You Need](https://mlsanity.com/paper/1706.03762/) (2017)
- [Teaching Machines to Read and Comprehend](https://mlsanity.com/paper/1506.03340/) (2015)
- [Bidirectional LSTM-CRF Models for Sequence Tagging](https://mlsanity.com/paper/1508.01991/) (2015)
- [Named Entity Recognition with Bidirectional LSTM-CNNs](https://mlsanity.com/paper/1511.08308/) (2015)
- [GECToR -- Grammatical Error Correction: Tag, Not Rewrite](https://mlsanity.com/paper/2005.12592/) (2020)
- [Parallel Iterative Edit Models for Local Sequence Transduction](https://mlsanity.com/paper/1910.02893/) (2019)
- [Cross-type Biomedical Named Entity Recognition with Deep Multi-Task Learning](https://mlsanity.com/paper/1801.09851/) (2018)
- [NeuroNER: an easy-to-use program for named-entity recognition based on neural networks](https://mlsanity.com/paper/1705.05487/) (2017)
- [Leveraging Text Repetitions and Denoising Autoencoders in OCR Post-correction](https://mlsanity.com/paper/1906.10907/) (2019)

## Built on by (most cited first, 11 of 11)

- [Synthetic Document Generator for Annotation-free Layout Recognition](https://mlsanity.com/paper/2111.06016/) (2021)
- [OCR Improves Machine Translation for Low-Resource Languages](https://mlsanity.com/paper/2202.13274/) (2022)
- [M3T: A New Benchmark Dataset for Multi-Modal Document-Level Machine Translation](https://mlsanity.com/paper/2406.08255/) (2024)
- [Reviving Cultural Heritage: A Novel Approach for Comprehensive Historical Document Restoration](https://mlsanity.com/paper/2507.05108/) (2025)
- [Reading the unreadable: Creating a dataset of 19th century English newspapers using image-to-text language models](https://mlsanity.com/paper/2502.14901/) (2025)
- [Revise: A Framework for Revising OCRed text in Practical Information Systems with Data Contamination Strategy](https://mlsanity.com/paper/2604.08115/) (2026)
- [Scrambled text: training Language Models to correct OCR errors using synthetic data](https://mlsanity.com/paper/2409.19735/) (2024)
- [FlexDoc: Parameterized Sampling for Diverse Multilingual Synthetic Documents for Training Document Understanding Models](https://mlsanity.com/paper/2510.02133/) (2025)
- [Data Generation for Post-OCR correction of Cyrillic handwriting](https://mlsanity.com/paper/2311.15896/) (2023)
- [Sim2Real Docs: Domain Randomization for Documents in Natural Scenes using Ray-traced Rendering](https://mlsanity.com/paper/2112.09220/) (2021)
- [Synth-JDoc: Synthesizing a Japanese Document Image Dataset for OCR with Diverse Layouts and Embedded Images](https://mlsanity.com/paper/2608.28248/) (2026)

## Abstract (arXiv)

Document digitization is essential for the digital transformation of our societies, yet a crucial step in the process, Optical Character Recognition (OCR), is still not perfect. Even commercial OCR systems can produce questionable output depending on the fidelity of the scanned documents. In this paper, we demonstrate an effective framework for mitigating OCR errors for any downstream NLP task, using Named Entity Recognition (NER) as an example. We first address the data scarcity problem for model training by constructing a document synthesis pipeline, generating realistic but degraded data with NER labels. We measure the NER accuracy drop at various degradation levels and show that a text restoration model, trained on the degraded data, significantly closes the NER accuracy gaps caused by OCR errors, including on an out-of-domain dataset. For the benefit of the community, we have made the document synthesis pipeline available as an open-source project.