{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2021:XEYJVJNBE4JLIHYGAE4AF6YLR7","short_pith_number":"pith:XEYJVJNB","schema_version":"1.0","canonical_sha256":"b9309aa5a12712b41f06013802fb0b8fe0501dcdd720883d79ee0e8729df7591","source":{"kind":"arxiv","id":"2112.09106","version":1},"attestation_state":"computed","paper":{"title":"RegionCLIP: Region-based Language-Image Pretraining","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":["cs.AI","cs.LG"],"primary_cat":"cs.CV","authors_text":"Chunyuan Li, Jianfeng Gao, Jianwei Yang, Liunian Harold Li, Luowei Zhou, Lu Yuan, Noel Codella, Pengchuan Zhang, Xiyang Dai, Yin Li, Yiwu Zhong","submitted_at":"2021-12-16T18:39:36Z","abstract_excerpt":"Contrastive language-image pretraining (CLIP) using image-text pairs has achieved impressive results on image classification in both zero-shot and transfer learning settings. However, we show that directly applying such models to recognize image regions for object detection leads to poor performance due to a domain shift: CLIP was trained to match an image as a whole to a text description, without capturing the fine-grained alignment between image regions and text spans. To mitigate this issue, we propose a new method called RegionCLIP that significantly extends CLIP to learn region-level visu"},"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":"2112.09106","kind":"arxiv","version":1},"metadata":{"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.CV","submitted_at":"2021-12-16T18:39:36Z","cross_cats_sorted":["cs.AI","cs.LG"],"title_canon_sha256":"f63f928dd6cc6c33bb07ace4bf90ff217a28652e66ed7c0feefe4ee167e49fd0","abstract_canon_sha256":"3bd7a42cccf3392d3b234bb411715aa0a8a65716c68ac3ed767ee8a3ce9e9998"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T03:41:35.960566Z","signature_b64":"PLNJW40SvIjBXFG+oD5dIk90rhMWoxV0iMs+8U3BmdmHBYEfKDM3kRE/DkmvjV9o4Y6/Ak34Nj/GHnDWIlYdDg==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"b9309aa5a12712b41f06013802fb0b8fe0501dcdd720883d79ee0e8729df7591","last_reissued_at":"2026-07-05T03:41:35.960158Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T03:41:35.960158Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"RegionCLIP: Region-based Language-Image Pretraining","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":["cs.AI","cs.LG"],"primary_cat":"cs.CV","authors_text":"Chunyuan Li, Jianfeng Gao, Jianwei Yang, Liunian Harold Li, Luowei Zhou, Lu Yuan, Noel Codella, Pengchuan Zhang, Xiyang Dai, Yin Li, Yiwu Zhong","submitted_at":"2021-12-16T18:39:36Z","abstract_excerpt":"Contrastive language-image pretraining (CLIP) using image-text pairs has achieved impressive results on image classification in both zero-shot and transfer learning settings. However, we show that directly applying such models to recognize image regions for object detection leads to poor performance due to a domain shift: CLIP was trained to match an image as a whole to a text description, without capturing the fine-grained alignment between image regions and text spans. To mitigate this issue, we propose a new method called RegionCLIP that significantly extends CLIP to learn region-level visu"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2112.09106","kind":"arxiv","version":1},"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/2112.09106/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":"2112.09106","created_at":"2026-07-05T03:41:35.960216+00:00"},{"alias_kind":"arxiv_version","alias_value":"2112.09106v1","created_at":"2026-07-05T03:41:35.960216+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2112.09106","created_at":"2026-07-05T03:41:35.960216+00:00"},{"alias_kind":"pith_short_12","alias_value":"XEYJVJNBE4JL","created_at":"2026-07-05T03:41:35.960216+00:00"},{"alias_kind":"pith_short_16","alias_value":"XEYJVJNBE4JLIHYG","created_at":"2026-07-05T03:41:35.960216+00:00"},{"alias_kind":"pith_short_8","alias_value":"XEYJVJNB","created_at":"2026-07-05T03:41:35.960216+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":2,"internal_anchor_count":0,"sample":[{"citing_arxiv_id":"2606.10769","citing_title":"ZODS-RS -- Zero-training Oriented Detection & Segmentation for Remote Sensing","ref_index":33,"is_internal_anchor":false},{"citing_arxiv_id":"2208.14649","citing_title":"DetailCLIP: Injecting Image Details into CLIP's Feature Space","ref_index":31,"is_internal_anchor":false}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/XEYJVJNBE4JLIHYGAE4AF6YLR7","json":"https://pith.science/pith/XEYJVJNBE4JLIHYGAE4AF6YLR7.json","graph_json":"https://pith.science/api/pith-number/XEYJVJNBE4JLIHYGAE4AF6YLR7/graph.json","events_json":"https://pith.science/api/pith-number/XEYJVJNBE4JLIHYGAE4AF6YLR7/events.json","paper":"https://pith.science/paper/XEYJVJNB"},"agent_actions":{"view_html":"https://pith.science/pith/XEYJVJNBE4JLIHYGAE4AF6YLR7","download_json":"https://pith.science/pith/XEYJVJNBE4JLIHYGAE4AF6YLR7.json","view_paper":"https://pith.science/paper/XEYJVJNB","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2112.09106&json=true","fetch_graph":"https://pith.science/api/pith-number/XEYJVJNBE4JLIHYGAE4AF6YLR7/graph.json","fetch_events":"https://pith.science/api/pith-number/XEYJVJNBE4JLIHYGAE4AF6YLR7/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/XEYJVJNBE4JLIHYGAE4AF6YLR7/action/timestamp_anchor","attest_storage":"https://pith.science/pith/XEYJVJNBE4JLIHYGAE4AF6YLR7/action/storage_attestation","attest_author":"https://pith.science/pith/XEYJVJNBE4JLIHYGAE4AF6YLR7/action/author_attestation","sign_citation":"https://pith.science/pith/XEYJVJNBE4JLIHYGAE4AF6YLR7/action/citation_signature","submit_replication":"https://pith.science/pith/XEYJVJNBE4JLIHYGAE4AF6YLR7/action/replication_record"}},"created_at":"2026-07-05T03:41:35.960216+00:00","updated_at":"2026-07-05T03:41:35.960216+00:00"}