{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2025:D3WOVQR4E3GSCDX7R7AU6V7JOT","short_pith_number":"pith:D3WOVQR4","schema_version":"1.0","canonical_sha256":"1eeceac23c26cd210eff8fc14f57e974db0cfd7394335448dd6c34117897517e","source":{"kind":"arxiv","id":"2504.02328","version":1},"attestation_state":"computed","paper":{"title":"Refining CLIP's Spatial Awareness: A Visual-Centric Perspective","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":[],"primary_cat":"cs.CV","authors_text":"Congpei Qiu, Tong Zhang, Wei Ke, Xiuxiu Bai, Yanhao Wu","submitted_at":"2025-04-03T07:04:56Z","abstract_excerpt":"Contrastive Language-Image Pre-training (CLIP) excels in global alignment with language but exhibits limited sensitivity to spatial information, leading to strong performance in zero-shot classification tasks but underperformance in tasks requiring precise spatial understanding. Recent approaches have introduced Region-Language Alignment (RLA) to enhance CLIP's performance in dense multimodal tasks by aligning regional visual representations with corresponding text inputs. However, we find that CLIP ViTs fine-tuned with RLA suffer from notable loss in spatial awareness, which is crucial for de"},"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":"2504.02328","kind":"arxiv","version":1},"metadata":{"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.CV","submitted_at":"2025-04-03T07:04:56Z","cross_cats_sorted":[],"title_canon_sha256":"7aaf7f2b7cbb3f45606c2b9a22e20df9e4ac023cf04438dc3ebda2ef3eeebb90","abstract_canon_sha256":"7741b732ac36e76627142621ed234173e336c7b7050ca400ec06f98601240c24"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T10:43:54.974766Z","signature_b64":"5pb6lCGGO+2kVxzxGtfCjAsbOgWwXuyg9CF2sNLfzJjWnaHXkO15Dr2RtqSxE7nhG31pbtI+HQ0YFZGWfyxwCQ==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"1eeceac23c26cd210eff8fc14f57e974db0cfd7394335448dd6c34117897517e","last_reissued_at":"2026-07-05T10:43:54.974282Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T10:43:54.974282Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"Refining CLIP's Spatial Awareness: A Visual-Centric Perspective","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":[],"primary_cat":"cs.CV","authors_text":"Congpei Qiu, Tong Zhang, Wei Ke, Xiuxiu Bai, Yanhao Wu","submitted_at":"2025-04-03T07:04:56Z","abstract_excerpt":"Contrastive Language-Image Pre-training (CLIP) excels in global alignment with language but exhibits limited sensitivity to spatial information, leading to strong performance in zero-shot classification tasks but underperformance in tasks requiring precise spatial understanding. Recent approaches have introduced Region-Language Alignment (RLA) to enhance CLIP's performance in dense multimodal tasks by aligning regional visual representations with corresponding text inputs. However, we find that CLIP ViTs fine-tuned with RLA suffer from notable loss in spatial awareness, which is crucial for de"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2504.02328","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/2504.02328/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":"2504.02328","created_at":"2026-07-05T10:43:54.974346+00:00"},{"alias_kind":"arxiv_version","alias_value":"2504.02328v1","created_at":"2026-07-05T10:43:54.974346+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2504.02328","created_at":"2026-07-05T10:43:54.974346+00:00"},{"alias_kind":"pith_short_12","alias_value":"D3WOVQR4E3GS","created_at":"2026-07-05T10:43:54.974346+00:00"},{"alias_kind":"pith_short_16","alias_value":"D3WOVQR4E3GSCDX7","created_at":"2026-07-05T10:43:54.974346+00:00"},{"alias_kind":"pith_short_8","alias_value":"D3WOVQR4","created_at":"2026-07-05T10:43:54.974346+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":3,"internal_anchor_count":1,"sample":[{"citing_arxiv_id":"2607.07135","citing_title":"Sparse Attention for Dense Open-Vocabulary Prediction in CLIP","ref_index":19,"is_internal_anchor":true},{"citing_arxiv_id":"2605.19622","citing_title":"UniRefiner: Teaching Pre-trained ViTs to Self-Dispose Dross via Contrastive Register","ref_index":24,"is_internal_anchor":false},{"citing_arxiv_id":"2601.01762","citing_title":"AlignDrive: Aligned Lateral-Longitudinal Planning for End-to-End Autonomous Driving","ref_index":7,"is_internal_anchor":false}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/D3WOVQR4E3GSCDX7R7AU6V7JOT","json":"https://pith.science/pith/D3WOVQR4E3GSCDX7R7AU6V7JOT.json","graph_json":"https://pith.science/api/pith-number/D3WOVQR4E3GSCDX7R7AU6V7JOT/graph.json","events_json":"https://pith.science/api/pith-number/D3WOVQR4E3GSCDX7R7AU6V7JOT/events.json","paper":"https://pith.science/paper/D3WOVQR4"},"agent_actions":{"view_html":"https://pith.science/pith/D3WOVQR4E3GSCDX7R7AU6V7JOT","download_json":"https://pith.science/pith/D3WOVQR4E3GSCDX7R7AU6V7JOT.json","view_paper":"https://pith.science/paper/D3WOVQR4","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2504.02328&json=true","fetch_graph":"https://pith.science/api/pith-number/D3WOVQR4E3GSCDX7R7AU6V7JOT/graph.json","fetch_events":"https://pith.science/api/pith-number/D3WOVQR4E3GSCDX7R7AU6V7JOT/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/D3WOVQR4E3GSCDX7R7AU6V7JOT/action/timestamp_anchor","attest_storage":"https://pith.science/pith/D3WOVQR4E3GSCDX7R7AU6V7JOT/action/storage_attestation","attest_author":"https://pith.science/pith/D3WOVQR4E3GSCDX7R7AU6V7JOT/action/author_attestation","sign_citation":"https://pith.science/pith/D3WOVQR4E3GSCDX7R7AU6V7JOT/action/citation_signature","submit_replication":"https://pith.science/pith/D3WOVQR4E3GSCDX7R7AU6V7JOT/action/replication_record"}},"created_at":"2026-07-05T10:43:54.974346+00:00","updated_at":"2026-07-05T10:43:54.974346+00:00"}