mlsanityData to 7 Oct 2026

Efficient OCR for Building a Diverse Digital History

arXiv 2304.02737 ↗First posted 5 Apr 2023cs.CV cs.DL econ.GNcited by 10

Not on the map. Why?

Abstract (arXiv)

Thousands of users consult digital archives daily, but the information they can access is unrepresentative of the diversity of documentary history. The sequence-to-sequence architecture typically used for optical character recognition (OCR) - which jointly learns a vision and language model - is poorly extensible to low-resource document collections, as learning a language-vision model requires extensive labeled sequences and compute. This study models OCR as a character level image retrieval problem, using a contrastively trained vision encoder. Because the model only learns characters' visual features, it is more sample efficient and extensible than existing architectures, enabling accurate OCR in settings where existing solutions fail. Crucially, the model opens new avenues for community engagement in making digital history more representative of documentary history.

Key facts. Cited by 10 papers in the map (data through 7 Oct 2026). Builds on 38 of its 64 references that are in the map. First posted on arXiv 5 Apr 2023.

Builds on

38 of 64 references.

References · 38

  1. 2021Swin Transformer: Hierarchical Vision Transformer using Shifted Windowscited by 12,082
  2. 2018Representation Learning with Contrastive Predictive Codingcited by 7,662
  3. 2017Mask R-CNNcited by 7,259
  4. 2020Bootstrap your own latent: A new approach to self-supervised Learningcited by 4,712
  5. 2022A ConvNet for the 2020scited by 3,545
  6. 2020Exploring Simple Siamese Representation Learningcited by 2,853
  7. 2020Supervised Contrastive Learningcited by 2,803
  8. 2017Billion-scale similarity search with GPUscited by 2,477
  9. 2019Searching for MobileNetV3cited by 1,984
  10. 2021An Empirical Study of Training Self-Supervised Vision Transformerscited by 1,610
  11. 2019MMDetection: Open MMLab Detection Toolbox and Benchmarkcited by 1,379
  12. 2020DocVQA: A Dataset for VQA on Document Imagescited by 1,288
26 more
  1. 2017Cascade R-CNN: Delving into High Quality Object Detectioncited by 1,275
  2. 2020The State and Fate of Linguistic Diversity and Inclusion in the NLP Worldcited by 789
  3. 2022Exploring Plain Vision Transformer Backbones for Object Detectioncited by 723
  4. 2015An End-to-End Trainable Neural Network for Image-based Sequence Recognition and Its Application to Scene Text Recognitioncited by 519
  5. 2019Cascade R-CNN: High Quality Object Detection and Instance Segmentationcited by 428
  6. 2021XCiT: Cross-Covariance Image Transformerscited by 335
  7. 2021TrOCR: Transformer-based Optical Character Recognition with Pre-trained Modelscited by 278
  8. 2021ICDAR2019 Competition on Scanned Receipt OCR and Information Extractioncited by 254
  9. 2021ReStyle: A Residual-Based StyleGAN Encoder via Iterative Refinementcited by 248
  10. 2022Splicing ViT Features for Semantic Appearance Transfercited by 209
  11. 2020A Survey on Recent Approaches for Natural Language Processing in Low-Resource Scenarioscited by 170
  12. 2021Training Vision Transformers for Image Retrievalcited by 90
  13. 2022SVTR: Scene Text Recognition with a Single Visual Modelcited by 86
  14. 2020Revisiting the Sibling Head in Object Detectorcited by 82
  15. 2020Sequence-to-Sequence Contrastive Learning for Text Recognitioncited by 70
  16. 2019Convolutional Character Networkscited by 61
  17. 2021LayoutParser: A Unified Toolkit for Deep Learning Based Document Image Analysiscited by 58
  18. 2020PyTorch Metric Learningcited by 57
  19. 2022Language Modelling with Pixelscited by 53
  20. 2021Handwriting Transformerscited by 37
  21. 2020OCR Post Correction for Endangered Language Textscited by 31
  22. 2019EATEN: Entity-aware Attention for Single Shot Visual Text Extractioncited by 29
  23. 2020A Large Dataset of Historical Japanese Documents with Complex Layoutscited by 24
  24. 2022A Survey of Historical Document Image Datasetscited by 18
  25. 2020The Newspaper Navigator Dataset: Extracting And Analyzing Visual Content from 16 Million Historic Newspaper Pages in Chronicling Americacited by 14
  26. 2021Neural OCR Post-Hoc Correction of Historical Corporacited by 12

Built on by

Cited by year (2026 to 7 Oct)

Citing papers per year: 2023: 4; 2024: 2; 2025: 3; 2026: 1.

10 papers.

  1. 2023American Stories: A Large-Scale Structured Text Dataset of Historical U.S. Newspaperscited by 11
  2. 2023EfficientOCR: An Extensible, Open-Source Package for Efficiently Digitizing World Knowledgecited by 5
  3. 2024Newswire: A Large-Scale Structured Database of a Century of Historical Newscited by 5
  4. 2024Deep Learning for Economistscited by 4
  5. 2025Reading the unreadable: Creating a dataset of 19th century English newspapers using image-to-text language modelscited by 3
  6. 2025DongbaMIE: A Multimodal Information Extraction Dataset for Evaluating Semantic Understanding of Dongba Pictogramscited by 3
  7. 2023Quantifying Character Similarity with Vision Transformerscited by 2
  8. 2023Linking Representations with Multimodal Contrastive Learningcited by 1
  9. 2025Layout-Aware OCR for Black Digital Archives with Unsupervised Evaluationcited by 1
  10. 2026Chronos: The AI Co-Historiancited by 0