{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2026:YNG5SRF7WQINTOTJ5OLUYD56NK","short_pith_number":"pith:YNG5SRF7","schema_version":"1.0","canonical_sha256":"c34dd944bfb410d9ba69eb974c0fbe6aa34ed6545cde1ce8d67bd1d41930942c","source":{"kind":"arxiv","id":"2605.08925","version":2},"attestation_state":"computed","paper":{"title":"ClickSeg3D: Few-Click Interactive Segmentation via Semantic Embeddings","license":"http://creativecommons.org/licenses/by/4.0/","headline":"A 3D interactive segmentation framework processes multiple user clicks together in one forward pass to label objects accurately.","cross_cats":[],"primary_cat":"cs.CV","authors_text":"Kourosh Khoshelham, Liangliang Nan, Xueyang Kang, Zijian Yu","submitted_at":"2026-05-09T12:51:25Z","abstract_excerpt":"Interactive segmentation allows efficient label generation by leveraging user-provided clicks to progressively refine predictions, which is critical when fully supervised labels are costly or generalization to unseen classes is needed. Existing 3D interactive methods are limited: most operate sequentially, predicting only one object per iteration with binary masks, while several recent approaches depend on 2D foundation models and camera alignment to bridge the 2D-3D gap. To address these limitations, we propose a novel interactive segmentation framework that operates directly on sparse, rando"},"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":true},"canonical_record":{"source":{"id":"2605.08925","kind":"arxiv","version":2},"metadata":{"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.CV","submitted_at":"2026-05-09T12:51:25Z","cross_cats_sorted":[],"title_canon_sha256":"453e5b49c2abd075224f8b51de49ed2ec5f9237903aecde9b985dfa70ebfea82","abstract_canon_sha256":"a053b98029fd6bb24df56678b3f3fb692a362fe496d3a879f48ca7cef668d8d5"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-05-20T00:03:15.537809Z","signature_b64":"mTltPLtABOW7NPBV9NRRK8jlFF/HoRPhRFQPBYIwTgEwsZgUOZNmPSJTG3twR5nYNge76oBGDojXRK/BqTcnAw==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"c34dd944bfb410d9ba69eb974c0fbe6aa34ed6545cde1ce8d67bd1d41930942c","last_reissued_at":"2026-05-20T00:03:15.536901Z","signature_status":"signed_v1","first_computed_at":"2026-05-20T00:03:15.536901Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"ClickSeg3D: Few-Click Interactive Segmentation via Semantic Embeddings","license":"http://creativecommons.org/licenses/by/4.0/","headline":"A 3D interactive segmentation framework processes multiple user clicks together in one forward pass to label objects accurately.","cross_cats":[],"primary_cat":"cs.CV","authors_text":"Kourosh Khoshelham, Liangliang Nan, Xueyang Kang, Zijian Yu","submitted_at":"2026-05-09T12:51:25Z","abstract_excerpt":"Interactive segmentation allows efficient label generation by leveraging user-provided clicks to progressively refine predictions, which is critical when fully supervised labels are costly or generalization to unseen classes is needed. Existing 3D interactive methods are limited: most operate sequentially, predicting only one object per iteration with binary masks, while several recent approaches depend on 2D foundation models and camera alignment to bridge the 2D-3D gap. To address these limitations, we propose a novel interactive segmentation framework that operates directly on sparse, rando"},"claims":{"count":4,"items":[{"kind":"strongest_claim","text":"Our framework ... processes multiple object clicks in a single forward pass. ... improves the mIoU metric by over 20 percent compared to strong baselines and achieves 8-10 percent gains under cross-dataset evaluation for a one-click per instance setting, often requiring only a single click per object.","source":"verdict.strongest_claim","status":"machine_extracted","claim_id":"C1","attestation":"unclaimed"},{"kind":"weakest_assumption","text":"That the hierarchical mask decoder with learnable semantic embeddings can jointly reason over all click queries, model inter-instance relationships, and refine both spatial masks and semantic predictions without requiring repeated model updates after each corrective click.","source":"verdict.weakest_assumption","status":"machine_extracted","claim_id":"C2","attestation":"unclaimed"},{"kind":"one_line_summary","text":"A point-Transformer interactive 3D instance segmentation model handles multiple clicks jointly in one pass and reports over 20% mIoU gains versus baselines plus 8-10% cross-dataset improvement for one-click-per-instance settings.","source":"verdict.one_line_summary","status":"machine_extracted","claim_id":"C3","attestation":"unclaimed"},{"kind":"headline","text":"A 3D interactive segmentation framework processes multiple user clicks together in one forward pass to label objects accurately.","source":"verdict.pith_extraction.headline","status":"machine_extracted","claim_id":"C4","attestation":"unclaimed"}],"snapshot_sha256":"1ce615e2ea45872e96806a3af013514c1d3c2e6a859b2f52d251d5953e05a0c1"},"source":{"id":"2605.08925","kind":"arxiv","version":2},"verdict":{"id":"08bb7332-fd0c-40c7-80fe-962f771e9a92","model_set":{"reader":"grok-4.3"},"created_at":"2026-05-12T01:46:05.620616Z","strongest_claim":"Our framework ... processes multiple object clicks in a single forward pass. ... improves the mIoU metric by over 20 percent compared to strong baselines and achieves 8-10 percent gains under cross-dataset evaluation for a one-click per instance setting, often requiring only a single click per object.","one_line_summary":"A point-Transformer interactive 3D instance segmentation model handles multiple clicks jointly in one pass and reports over 20% mIoU gains versus baselines plus 8-10% cross-dataset improvement for one-click-per-instance settings.","pipeline_version":"pith-pipeline@v0.9.0","weakest_assumption":"That the hierarchical mask decoder with learnable semantic embeddings can jointly reason over all click queries, model inter-instance relationships, and refine both spatial masks and semantic predictions without requiring repeated model updates after each corrective click.","pith_extraction_headline":"A 3D interactive segmentation framework processes multiple user clicks together in one forward pass to label objects accurately."},"integrity":{"clean":true,"summary":{"advisory":0,"critical":0,"by_detector":{},"informational":0},"endpoint":"/pith/2605.08925/integrity.json","findings":[],"available":true,"detectors_run":[{"name":"ai_meta_artifact","ran_at":"2026-05-19T20:41:12.751603Z","status":"completed","version":"1.0.0","findings_count":0},{"name":"doi_title_agreement","ran_at":"2026-05-19T14:01:19.831685Z","status":"completed","version":"1.0.0","findings_count":0},{"name":"doi_compliance","ran_at":"2026-05-19T10:40:50.241599Z","status":"completed","version":"1.0.0","findings_count":0}],"snapshot_sha256":"95f1a3faac8ab6a88f7b3cf4332139ac93ef4eca7da140682ba194eeaf424da2"},"references":{"count":0,"sample":[],"resolved_work":0,"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57","internal_anchors":0},"formal_canon":{"evidence_count":2,"snapshot_sha256":"bf50a21a4a425769a7f2ec72de9b9d2e3711267e1890a2f291ee72dd74fbbb30"},"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":"2605.08925","created_at":"2026-05-20T00:03:15.537048+00:00"},{"alias_kind":"arxiv_version","alias_value":"2605.08925v2","created_at":"2026-05-20T00:03:15.537048+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2605.08925","created_at":"2026-05-20T00:03:15.537048+00:00"},{"alias_kind":"pith_short_12","alias_value":"YNG5SRF7WQIN","created_at":"2026-05-20T00:03:15.537048+00:00"},{"alias_kind":"pith_short_16","alias_value":"YNG5SRF7WQINTOTJ","created_at":"2026-05-20T00:03:15.537048+00:00"},{"alias_kind":"pith_short_8","alias_value":"YNG5SRF7","created_at":"2026-05-20T00:03:15.537048+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":0,"internal_anchor_count":0,"sample":[]},"formal_canon":{"evidence_count":2,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/YNG5SRF7WQINTOTJ5OLUYD56NK","json":"https://pith.science/pith/YNG5SRF7WQINTOTJ5OLUYD56NK.json","graph_json":"https://pith.science/api/pith-number/YNG5SRF7WQINTOTJ5OLUYD56NK/graph.json","events_json":"https://pith.science/api/pith-number/YNG5SRF7WQINTOTJ5OLUYD56NK/events.json","paper":"https://pith.science/paper/YNG5SRF7"},"agent_actions":{"view_html":"https://pith.science/pith/YNG5SRF7WQINTOTJ5OLUYD56NK","download_json":"https://pith.science/pith/YNG5SRF7WQINTOTJ5OLUYD56NK.json","view_paper":"https://pith.science/paper/YNG5SRF7","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2605.08925&json=true","fetch_graph":"https://pith.science/api/pith-number/YNG5SRF7WQINTOTJ5OLUYD56NK/graph.json","fetch_events":"https://pith.science/api/pith-number/YNG5SRF7WQINTOTJ5OLUYD56NK/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/YNG5SRF7WQINTOTJ5OLUYD56NK/action/timestamp_anchor","attest_storage":"https://pith.science/pith/YNG5SRF7WQINTOTJ5OLUYD56NK/action/storage_attestation","attest_author":"https://pith.science/pith/YNG5SRF7WQINTOTJ5OLUYD56NK/action/author_attestation","sign_citation":"https://pith.science/pith/YNG5SRF7WQINTOTJ5OLUYD56NK/action/citation_signature","submit_replication":"https://pith.science/pith/YNG5SRF7WQINTOTJ5OLUYD56NK/action/replication_record"}},"created_at":"2026-05-20T00:03:15.537048+00:00","updated_at":"2026-05-20T00:03:15.537048+00:00"}