{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2021:QXJI5MUPVQ4XNNRWFXBOVSE33D","short_pith_number":"pith:QXJI5MUP","schema_version":"1.0","canonical_sha256":"85d28eb28fac3976b6362dc2eac89bd8eeabfa9a77e16aec9c80c5adaf4e6e5f","source":{"kind":"arxiv","id":"2109.10282","version":5},"attestation_state":"computed","paper":{"title":"TrOCR: Transformer-based Optical Character Recognition with Pre-trained Models","license":"http://creativecommons.org/licenses/by-nc-sa/4.0/","headline":"","cross_cats":["cs.CV"],"primary_cat":"cs.CL","authors_text":"Cha Zhang, Dinei Florencio, Furu Wei, Jingye Chen, Lei Cui, Minghao Li, Tengchao Lv, Yijuan Lu, Zhoujun Li","submitted_at":"2021-09-21T16:01:56Z","abstract_excerpt":"Text recognition is a long-standing research problem for document digitalization. Existing approaches are usually built based on CNN for image understanding and RNN for char-level text generation. In addition, another language model is usually needed to improve the overall accuracy as a post-processing step. In this paper, we propose an end-to-end text recognition approach with pre-trained image Transformer and text Transformer models, namely TrOCR, which leverages the Transformer architecture for both image understanding and wordpiece-level text generation. The TrOCR model is simple but effec"},"verification_status":{"content_addressed":true,"pith_receipt":true,"author_attested":false,"weak_author_claims":0,"strong_author_claims":0,"externally_anchored":false,"storage_verified":false,"citation_signatures":0,"replication_records":0,"graph_snapshot":true,"references_resolved":false,"formal_links_present":false},"canonical_record":{"source":{"id":"2109.10282","kind":"arxiv","version":5},"metadata":{"license":"http://creativecommons.org/licenses/by-nc-sa/4.0/","primary_cat":"cs.CL","submitted_at":"2021-09-21T16:01:56Z","cross_cats_sorted":["cs.CV"],"title_canon_sha256":"01b0666ea9d4bad14d14a476e9d9196fbb042fb6dcd8d6a7ebf2734123cf85ef","abstract_canon_sha256":"7e1b933274d74ed3d9d68d4b2a769e7f256c3525e87903bac4ba8a1dc52ac96a"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T04:54:46.881403Z","signature_b64":"Q4ihAOEcKzBg3mWAzUAR6MKlf0qdYIbR1DzlNS3XEPL1keEyaD0nRt7I4ZNiREo470q2iTwNXyiczfbbT8yXAg==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"85d28eb28fac3976b6362dc2eac89bd8eeabfa9a77e16aec9c80c5adaf4e6e5f","last_reissued_at":"2026-07-05T04:54:46.880982Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T04:54:46.880982Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"TrOCR: Transformer-based Optical Character Recognition with Pre-trained Models","license":"http://creativecommons.org/licenses/by-nc-sa/4.0/","headline":"","cross_cats":["cs.CV"],"primary_cat":"cs.CL","authors_text":"Cha Zhang, Dinei Florencio, Furu Wei, Jingye Chen, Lei Cui, Minghao Li, Tengchao Lv, Yijuan Lu, Zhoujun Li","submitted_at":"2021-09-21T16:01:56Z","abstract_excerpt":"Text recognition is a long-standing research problem for document digitalization. Existing approaches are usually built based on CNN for image understanding and RNN for char-level text generation. In addition, another language model is usually needed to improve the overall accuracy as a post-processing step. In this paper, we propose an end-to-end text recognition approach with pre-trained image Transformer and text Transformer models, namely TrOCR, which leverages the Transformer architecture for both image understanding and wordpiece-level text generation. The TrOCR model is simple but effec"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2109.10282","kind":"arxiv","version":5},"verdict":{"id":null,"model_set":{},"created_at":null,"strongest_claim":"","one_line_summary":"","pipeline_version":null,"weakest_assumption":"","pith_extraction_headline":""},"integrity":{"clean":true,"summary":{"advisory":0,"critical":0,"by_detector":{},"informational":0},"endpoint":"/pith/2109.10282/integrity.json","findings":[],"available":true,"detectors_run":[],"snapshot_sha256":"c28c3603d3b5d939e8dc4c7e95fa8dfce3d595e45f758748cecf8e644a296938"},"references":{"count":0,"sample":[],"resolved_work":0,"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57","internal_anchors":0},"formal_canon":{"evidence_count":0,"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"author_claims":{"count":0,"strong_count":0,"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"builder_version":"pith-number-builder-2026-05-17-v1"},"aliases":[{"alias_kind":"arxiv","alias_value":"2109.10282","created_at":"2026-07-05T04:54:46.881048+00:00"},{"alias_kind":"arxiv_version","alias_value":"2109.10282v5","created_at":"2026-07-05T04:54:46.881048+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2109.10282","created_at":"2026-07-05T04:54:46.881048+00:00"},{"alias_kind":"pith_short_12","alias_value":"QXJI5MUPVQ4X","created_at":"2026-07-05T04:54:46.881048+00:00"},{"alias_kind":"pith_short_16","alias_value":"QXJI5MUPVQ4XNNRW","created_at":"2026-07-05T04:54:46.881048+00:00"},{"alias_kind":"pith_short_8","alias_value":"QXJI5MUP","created_at":"2026-07-05T04:54:46.881048+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":10,"internal_anchor_count":0,"sample":[{"citing_arxiv_id":"2606.23144","citing_title":"Koshur Pixel: a large-scale synthetic ocr dataset for kashmiri","ref_index":28,"is_internal_anchor":false},{"citing_arxiv_id":"2605.27467","citing_title":"Comparative Analysis of Liquid Neural Networks and LSTM for Sequential Pattern Recognition: Robustness, Efficiency, and Clinical Utility","ref_index":27,"is_internal_anchor":false},{"citing_arxiv_id":"2304.03427","citing_title":"Cleansing Jewel: A Neural Spelling Correction Model Built On Google OCR-ed Tibetan Manuscripts","ref_index":7,"is_internal_anchor":false},{"citing_arxiv_id":"2511.02014","citing_title":"Towards Selection of Large Multimodal Models as Engines for Burned-in Protected Health Information Detection in Medical Images","ref_index":21,"is_internal_anchor":false},{"citing_arxiv_id":"2308.13418","citing_title":"Nougat: Neural Optical Understanding for Academic Documents","ref_index":8,"is_internal_anchor":false},{"citing_arxiv_id":"2602.01785","citing_title":"CodeOCR: On the Effectiveness of Vision Language Models in Code Understanding","ref_index":57,"is_internal_anchor":false},{"citing_arxiv_id":"2604.23644","citing_title":"RaV-IDP: A Reconstruction-as-Validation Framework for Faithful Intelligent Document Processing","ref_index":7,"is_internal_anchor":false},{"citing_arxiv_id":"2604.16504","citing_title":"From Handwriting to Structured Data: Benchmarking AI Digitisation of Handwritten Forms","ref_index":4,"is_internal_anchor":false},{"citing_arxiv_id":"2604.06160","citing_title":"The Character Error Vector: Decomposable errors for page-level OCR evaluation","ref_index":23,"is_internal_anchor":false},{"citing_arxiv_id":"2303.03378","citing_title":"PaLM-E: An Embodied Multimodal Language Model","ref_index":21,"is_internal_anchor":false}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/QXJI5MUPVQ4XNNRWFXBOVSE33D","json":"https://pith.science/pith/QXJI5MUPVQ4XNNRWFXBOVSE33D.json","graph_json":"https://pith.science/api/pith-number/QXJI5MUPVQ4XNNRWFXBOVSE33D/graph.json","events_json":"https://pith.science/api/pith-number/QXJI5MUPVQ4XNNRWFXBOVSE33D/events.json","paper":"https://pith.science/paper/QXJI5MUP"},"agent_actions":{"view_html":"https://pith.science/pith/QXJI5MUPVQ4XNNRWFXBOVSE33D","download_json":"https://pith.science/pith/QXJI5MUPVQ4XNNRWFXBOVSE33D.json","view_paper":"https://pith.science/paper/QXJI5MUP","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2109.10282&json=true","fetch_graph":"https://pith.science/api/pith-number/QXJI5MUPVQ4XNNRWFXBOVSE33D/graph.json","fetch_events":"https://pith.science/api/pith-number/QXJI5MUPVQ4XNNRWFXBOVSE33D/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/QXJI5MUPVQ4XNNRWFXBOVSE33D/action/timestamp_anchor","attest_storage":"https://pith.science/pith/QXJI5MUPVQ4XNNRWFXBOVSE33D/action/storage_attestation","attest_author":"https://pith.science/pith/QXJI5MUPVQ4XNNRWFXBOVSE33D/action/author_attestation","sign_citation":"https://pith.science/pith/QXJI5MUPVQ4XNNRWFXBOVSE33D/action/citation_signature","submit_replication":"https://pith.science/pith/QXJI5MUPVQ4XNNRWFXBOVSE33D/action/replication_record"}},"created_at":"2026-07-05T04:54:46.881048+00:00","updated_at":"2026-07-05T04:54:46.881048+00:00"}