{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2022:XIWPBS6OPPZFI3267L2HVLSEGE","short_pith_number":"pith:XIWPBS6O","schema_version":"1.0","canonical_sha256":"ba2cf0cbce7bf2546f5efaf47aae443120419e24145cdf0185efd98a554b879c","source":{"kind":"arxiv","id":"2209.05534","version":3},"attestation_state":"computed","paper":{"title":"PreSTU: Pre-Training for Scene-Text Understanding","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":["cs.CL"],"primary_cat":"cs.CV","authors_text":"Hexiang Hu, Jihyung Kil, Radu Soricut, Sebastian Goodman, Soravit Changpinyo, Wei-Lun Chao, Xi Chen","submitted_at":"2022-09-12T18:29:55Z","abstract_excerpt":"The ability to recognize and reason about text embedded in visual inputs is often lacking in vision-and-language (V&L) models, perhaps because V&L pre-training methods have often failed to include such an ability in their training objective. In this paper, we propose PreSTU, a novel pre-training recipe dedicated to scene-text understanding (STU). PreSTU introduces OCR-aware pre-training objectives that encourage the model to recognize text from an image and connect it to the rest of the image content. We implement PreSTU using a simple transformer-based encoder-decoder architecture, combined w"},"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":"2209.05534","kind":"arxiv","version":3},"metadata":{"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.CV","submitted_at":"2022-09-12T18:29:55Z","cross_cats_sorted":["cs.CL"],"title_canon_sha256":"4992ff540898dd4a9b7535079b4f069090cd34ca31ae4f89d9afc8c74227e09a","abstract_canon_sha256":"53a5935f5b61bf6ae786590f67af3c55f973e02b7260843d0ca7334dbd66369f"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T06:42:40.075655Z","signature_b64":"6uf43EfT4WZhaIIvUaF/p0X61iA3jRMiJu8eZkiKw9LpS06hEuf76IaaXO9/cOZ3rybY+TaM16vv7fdOZCe8Ag==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"ba2cf0cbce7bf2546f5efaf47aae443120419e24145cdf0185efd98a554b879c","last_reissued_at":"2026-07-05T06:42:40.075172Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T06:42:40.075172Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"PreSTU: Pre-Training for Scene-Text Understanding","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":["cs.CL"],"primary_cat":"cs.CV","authors_text":"Hexiang Hu, Jihyung Kil, Radu Soricut, Sebastian Goodman, Soravit Changpinyo, Wei-Lun Chao, Xi Chen","submitted_at":"2022-09-12T18:29:55Z","abstract_excerpt":"The ability to recognize and reason about text embedded in visual inputs is often lacking in vision-and-language (V&L) models, perhaps because V&L pre-training methods have often failed to include such an ability in their training objective. In this paper, we propose PreSTU, a novel pre-training recipe dedicated to scene-text understanding (STU). PreSTU introduces OCR-aware pre-training objectives that encourage the model to recognize text from an image and connect it to the rest of the image content. We implement PreSTU using a simple transformer-based encoder-decoder architecture, combined w"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2209.05534","kind":"arxiv","version":3},"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/2209.05534/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":"2209.05534","created_at":"2026-07-05T06:42:40.075225+00:00"},{"alias_kind":"arxiv_version","alias_value":"2209.05534v3","created_at":"2026-07-05T06:42:40.075225+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2209.05534","created_at":"2026-07-05T06:42:40.075225+00:00"},{"alias_kind":"pith_short_12","alias_value":"XIWPBS6OPPZF","created_at":"2026-07-05T06:42:40.075225+00:00"},{"alias_kind":"pith_short_16","alias_value":"XIWPBS6OPPZFI326","created_at":"2026-07-05T06:42:40.075225+00:00"},{"alias_kind":"pith_short_8","alias_value":"XIWPBS6O","created_at":"2026-07-05T06:42:40.075225+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":2,"internal_anchor_count":0,"sample":[{"citing_arxiv_id":"2305.18565","citing_title":"PaLI-X: On Scaling up a Multilingual Vision and Language Model","ref_index":24,"is_internal_anchor":false},{"citing_arxiv_id":"2209.06794","citing_title":"PaLI: A Jointly-Scaled Multilingual Language-Image Model","ref_index":45,"is_internal_anchor":false}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/XIWPBS6OPPZFI3267L2HVLSEGE","json":"https://pith.science/pith/XIWPBS6OPPZFI3267L2HVLSEGE.json","graph_json":"https://pith.science/api/pith-number/XIWPBS6OPPZFI3267L2HVLSEGE/graph.json","events_json":"https://pith.science/api/pith-number/XIWPBS6OPPZFI3267L2HVLSEGE/events.json","paper":"https://pith.science/paper/XIWPBS6O"},"agent_actions":{"view_html":"https://pith.science/pith/XIWPBS6OPPZFI3267L2HVLSEGE","download_json":"https://pith.science/pith/XIWPBS6OPPZFI3267L2HVLSEGE.json","view_paper":"https://pith.science/paper/XIWPBS6O","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2209.05534&json=true","fetch_graph":"https://pith.science/api/pith-number/XIWPBS6OPPZFI3267L2HVLSEGE/graph.json","fetch_events":"https://pith.science/api/pith-number/XIWPBS6OPPZFI3267L2HVLSEGE/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/XIWPBS6OPPZFI3267L2HVLSEGE/action/timestamp_anchor","attest_storage":"https://pith.science/pith/XIWPBS6OPPZFI3267L2HVLSEGE/action/storage_attestation","attest_author":"https://pith.science/pith/XIWPBS6OPPZFI3267L2HVLSEGE/action/author_attestation","sign_citation":"https://pith.science/pith/XIWPBS6OPPZFI3267L2HVLSEGE/action/citation_signature","submit_replication":"https://pith.science/pith/XIWPBS6OPPZFI3267L2HVLSEGE/action/replication_record"}},"created_at":"2026-07-05T06:42:40.075225+00:00","updated_at":"2026-07-05T06:42:40.075225+00:00"}