{"url":"/sota/handwritten-text-recognition-on-iam","task":{"name":"Handwritten Text Recognition","url":"/task/handwritten-text-recognition","note":null},"dataset":{"name":"IAM","url":"/dataset/iam"},"category":"Computer Vision","categories":["Adversarial","Computer Vision"],"category_note":null,"description":"Handwritten Text Recognition (HTR) is the task of automatically identifying and transcribing handwritten text from images or scanned documents into machine-readable text. The goal is to develop a system capable of accurately interpreting diverse handwriting styles, accounting for variations in alignment, stroke, spacing, and noise. This task involves detecting handwritten regions within an image, extracting the text content, and converting it into a structured digital format, enabling further search, indexing, or data analysis.","description_from":"task","source":{"archive":"pwc-archive (Hugging Face), CC BY-SA 4.0","snapshot":"2025-07-28","rank":"the archive's row order at snapshot; not re-ranked","rows_end_at":"2025-07-28","rows_withheld_as_spam":0,"metric_values":"the archive's strings, untouched"},"metrics":["CER","WER"],"metric_direction":{"note":"inferred from the metric name only (the archive records no direction); null = not inferred, chart draws points only","by_metric":{"CER":"lower","WER":"lower"}},"counts":{"rows":17,"rows_with_code":13,"rows_with_paper_page":17,"rows_dated":17,"rows_using_additional_data":3},"rows":[{"rank_in_archive_order":1,"model":"DTrOCR 105M","metrics":{"CER":"2.38"},"uses_additional_data":false,"paper_date":"2023-08-30","paper":"/paper/dtrocr-decoder-only-transformer-for-optical","paper_url":"https://arxiv.org/abs/2308.15996v1","paper_title":"DTrOCR: Decoder-only Transformer for Optical Character Recognition","code":"https://github.com/arvindrajan92/DTrOCR","n_code_links":1,"syntology":null},{"rank_in_archive_order":2,"model":"Self-Attention + CTC + language model","metrics":{"CER":"2.75"},"uses_additional_data":true,"paper_date":"2021-04-15","paper":"/paper/rethinking-text-line-recognition-models","paper_url":"https://arxiv.org/abs/2104.07787v2","paper_title":"Rethinking Text Line Recognition Models","code":null,"n_code_links":0,"syntology":null},{"rank_in_archive_order":3,"model":"TrOCR-large 558M","metrics":{"CER":"2.89"},"uses_additional_data":false,"paper_date":"2021-09-21","paper":"/paper/trocr-transformer-based-optical-character","paper_url":"https://arxiv.org/abs/2109.10282v5","paper_title":"TrOCR: Transformer-based Optical Character Recognition with Pre-trained Models","code":"https://github.com/huggingface/transformers","n_code_links":8,"syntology":{"n_ran":0,"n_unverified":6,"n_samples":6,"n_pointer_only_licence":0}},{"rank_in_archive_order":4,"model":"Transformer + CNN","metrics":{"CER":"2.96"},"uses_additional_data":true,"paper_date":"2021-04-15","paper":"/paper/rethinking-text-line-recognition-models","paper_url":"https://arxiv.org/abs/2104.07787v2","paper_title":"Rethinking Text Line Recognition Models","code":null,"n_code_links":0,"syntology":null},{"rank_in_archive_order":5,"model":"TrOCR-base 334M","metrics":{"CER":"3.42"},"uses_additional_data":false,"paper_date":"2021-09-21","paper":"/paper/trocr-transformer-based-optical-character","paper_url":"https://arxiv.org/abs/2109.10282v5","paper_title":"TrOCR: Transformer-based Optical Character Recognition with Pre-trained Models","code":"https://github.com/huggingface/transformers","n_code_links":8,"syntology":{"n_ran":0,"n_unverified":6,"n_samples":6,"n_pointer_only_licence":0}},{"rank_in_archive_order":6,"model":"TrOCR-small 62M","metrics":{"CER":"4.22"},"uses_additional_data":false,"paper_date":"2021-09-21","paper":"/paper/trocr-transformer-based-optical-character","paper_url":"https://arxiv.org/abs/2109.10282v5","paper_title":"TrOCR: Transformer-based Optical Character Recognition with Pre-trained Models","code":"https://github.com/huggingface/transformers","n_code_links":8,"syntology":{"n_ran":0,"n_unverified":6,"n_samples":6,"n_pointer_only_licence":0}},{"rank_in_archive_order":7,"model":"VAN","metrics":{"CER":"4.32","WER":"16.24"},"uses_additional_data":false,"paper_date":"2020-12-07","paper":"/paper/end-to-end-handwritten-paragraph-text","paper_url":"https://arxiv.org/abs/2012.03868v2","paper_title":"End-to-end Handwritten Paragraph Text Recognition Using a Vertical Attention Network","code":"https://github.com/FactoDeepLearning/VerticalAttentionOCR","n_code_links":1,"syntology":null},{"rank_in_archive_order":8,"model":"Transformer w/ CNN (+synth)","metrics":{"CER":"4.67"},"uses_additional_data":false,"paper_date":"2020-05-26","paper":"/paper/pay-attention-to-what-you-read-non-recurrent","paper_url":"https://arxiv.org/abs/2005.13044v1","paper_title":"Pay Attention to What You Read: Non-recurrent Handwritten Text-Line Recognition","code":null,"n_code_links":0,"syntology":null},{"rank_in_archive_order":9,"model":"HTR-VT(line-level)","metrics":{"CER":"4.7","WER":"14.9"},"uses_additional_data":false,"paper_date":"2024-09-13","paper":"/paper/htr-vt-handwritten-text-recognition-with","paper_url":"https://arxiv.org/abs/2409.08573v1","paper_title":"HTR-VT: Handwritten Text Recognition with Vision Transformer","code":"https://github.com/yutingli0606/htr-vt","n_code_links":2,"syntology":null},{"rank_in_archive_order":10,"model":"Easter2.0","metrics":{"CER":"6.21"},"uses_additional_data":true,"paper_date":"2022-05-30","paper":"/paper/easter2-0-improving-convolutional-models-for","paper_url":"https://arxiv.org/abs/2205.14879v1","paper_title":"Easter2.0: Improving convolutional models for handwritten text recognition","code":"https://github.com/kartikgill/easter2","n_code_links":1,"syntology":null},{"rank_in_archive_order":11,"model":"FPHR+Aug Paragraph Level (~145 dpi)","metrics":{"CER":"6.3"},"uses_additional_data":false,"paper_date":"2021-03-11","paper":"/paper/full-page-handwriting-recognition-via-image","paper_url":"https://arxiv.org/abs/2103.06450v3","paper_title":"Full Page Handwriting Recognition via Image to Sequence Extraction","code":"https://github.com/kingyiusuen/image-to-latex","n_code_links":5,"syntology":null},{"rank_in_archive_order":12,"model":"Decouple Attention Network","metrics":{"CER":"6.4","WER":"19.6"},"uses_additional_data":false,"paper_date":"2019-12-21","paper":"/paper/decoupled-attention-network-for-text","paper_url":"https://arxiv.org/abs/1912.10205v1","paper_title":"Decoupled Attention Network for Text Recognition","code":"https://github.com/topdu/openocr","n_code_links":5,"syntology":null},{"rank_in_archive_order":13,"model":"Start, Follow, Read","metrics":{"CER":"6.4","WER":"23.2"},"uses_additional_data":false,"paper_date":"2018-09-01","paper":"/paper/start-follow-read-end-to-end-full-page","paper_url":"http://openaccess.thecvf.com/content_ECCV_2018/html/Curtis_Wigington_Start_Follow_Read_ECCV_2018_paper.html","paper_title":"Start, Follow, Read: End-to-End Full-Page Handwriting Recognition","code":"https://github.com/cwig/start_follow_read","n_code_links":1,"syntology":null},{"rank_in_archive_order":14,"model":"FPHR+Aug Line Level (~145 dpi)","metrics":{"CER":"6.5"},"uses_additional_data":false,"paper_date":"2021-03-11","paper":"/paper/full-page-handwriting-recognition-via-image","paper_url":"https://arxiv.org/abs/2103.06450v3","paper_title":"Full Page Handwriting Recognition via Image to Sequence Extraction","code":"https://github.com/kingyiusuen/image-to-latex","n_code_links":5,"syntology":null},{"rank_in_archive_order":15,"model":"Leaky LP Cell","metrics":{"CER":"6.6","WER":"15.9"},"uses_additional_data":false,"paper_date":"2019-02-28","paper":"/paper/no-padding-please-efficient-neural","paper_url":"http://arxiv.org/abs/1902.11208v1","paper_title":"No Padding Please: Efficient Neural Handwriting Recognition","code":"https://github.com/gwenniger/multi-hare","n_code_links":1,"syntology":null},{"rank_in_archive_order":16,"model":"FPHR Paragraph Level (~145 dpi)","metrics":{"CER":"6.7"},"uses_additional_data":false,"paper_date":"2021-03-11","paper":"/paper/full-page-handwriting-recognition-via-image","paper_url":"https://arxiv.org/abs/2103.06450v3","paper_title":"Full Page Handwriting Recognition via Image to Sequence Extraction","code":"https://github.com/kingyiusuen/image-to-latex","n_code_links":5,"syntology":null},{"rank_in_archive_order":17,"model":"Transformer w/ CNN","metrics":{"CER":"7.62"},"uses_additional_data":false,"paper_date":"2020-05-26","paper":"/paper/pay-attention-to-what-you-read-non-recurrent","paper_url":"https://arxiv.org/abs/2005.13044v1","paper_title":"Pay Attention to What You Read: Non-recurrent Handwritten Text-Line Recognition","code":null,"n_code_links":0,"syntology":null}],"since_archive":{"claim":"Results that newer papers report for their own method, placed here by Syntology. A model pointed at the cell in the paper's own table; the number was read from that cell and checked against this leaderboard's metric, dataset, split and scale; an independent check that saw this leaderboard's other rows and every other leaderboard on the same dataset accepted it. Not reviewed by the paper's authors or by the archive's editors, and not ranked against the archive rows.","extraction_file_present":true,"measurement":{"test_papers":883,"papers_with_output":881,"judged_true":108,"judged":110,"wilson95_lower":0.9361,"measured_on":"2026-09-24","frozen_commit":"0e3de0df94"},"measurement_note":"blind adjudication of accepted entries on a held-out split of archive papers, rules frozen before the test","coverage":{"sentence":"Syntology has checked 6,264 of the 9,581 papers on this site that are newer than the archive; results from the others appear after they are checked.","complete":false,"papers_newer_than_archive":9581,"papers_checked":6264,"papers_extracted_not_yet_verified":0,"boards_without_verdict":2,"papers_not_yet_extracted":3316},"order":"newest first by month (arXiv date, else the arXiv-id month), then arXiv id descending","columns":[],"entries":[]},"syntology":{"read_at":"2026-09-24T18:15:14+00:00","claim":"Per row: N of M harvested code samples from that row's paper executed on a synthesized fixture; the other M-N are unverified. Not a reproduction of the row's number; not a correctness claim. n_pointer_only_licence counts samples the site points at rather than redistributes (a licence axis, independent of ran/unverified).","rows_with_graph_line":3,"rows_with_any_sample_ran":0,"distinct_papers_with_graph_line":1,"distinct_papers_with_any_sample_ran":0,"samples_over_distinct_papers":{"n_ran":0,"n_unverified":6,"n_samples":6,"n_pointer_only_licence":0,"note":"each paper (arXiv id) counted once, however many rows it is behind; this is the page-level figure"},"samples_row_weighted":{"n_ran":0,"n_unverified":18,"n_samples":18,"n_pointer_only_licence":0,"note":"row-weighted: a paper behind several rows is counted once per row; inflated relative to samples_over_distinct_papers by design, kept for readers summing the per-row syntology blocks"}}}