{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2026:NMXAYK3NTGOHEIQIE6UPZMNQHO","short_pith_number":"pith:NMXAYK3N","schema_version":"1.0","canonical_sha256":"6b2e0c2b6d999c72220827a8fcb1b03ba3110ed4cd1b6353c43a648ddcf241ad","source":{"kind":"arxiv","id":"2604.12918","version":2},"attestation_state":"computed","paper":{"title":"Radar-Camera BEV Multi-Task Learning with Cross-Task Attention Bridge for Joint 3D Detection and Segmentation","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"A bidirectional attention bridge in shared BEV space lets detection and segmentation exchange features, raising segmentation accuracy on seven classes while detection stays neutral.","cross_cats":[],"primary_cat":"cs.CV","authors_text":"Ahmet \\.Inan\\c{c}, \\\"Ozg\\\"ur Erkent","submitted_at":"2026-04-14T16:00:53Z","abstract_excerpt":"Bird's-eye-view (BEV) representations are the dominant paradigm for 3D perception in autonomous driving, providing a unified spatial canvas where detection and segmentation features are geometrically registered to the same physical coordinate system. However, existing radar-camera fusion methods treat these tasks in isolation, missing the opportunity for cross-task feature sharing: object-level geometric cues from detection can sharpen segmentation, while dense road-layout context from segmentation can anchor detection. We propose \\textbf{CTAB} (Cross-Task Attention Bridge), a bidirectional mo"},"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":"2604.12918","kind":"arxiv","version":2},"metadata":{"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.CV","submitted_at":"2026-04-14T16:00:53Z","cross_cats_sorted":[],"title_canon_sha256":"30ecc991678a7bbf4e1f7a2ef0bcb08c329a8c309b533be728123c1ea8ef161d","abstract_canon_sha256":"3a1cfa7ea5ac46f875e8f0b3f3a4f9c2559ef902b07ee8211fdb590f8da3e268"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-05-27T02:06:14.051878Z","signature_b64":"fXYnsFqzBRxnUoJT5T919+Afp/OcOK/YyiU3kn6jLX/MleJi5VxO05w6xcmlVYTLR3dzUfk2Jvrmzym+RPsqCg==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"6b2e0c2b6d999c72220827a8fcb1b03ba3110ed4cd1b6353c43a648ddcf241ad","last_reissued_at":"2026-05-27T02:06:14.051020Z","signature_status":"signed_v1","first_computed_at":"2026-05-27T02:06:14.051020Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"Radar-Camera BEV Multi-Task Learning with Cross-Task Attention Bridge for Joint 3D Detection and Segmentation","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"A bidirectional attention bridge in shared BEV space lets detection and segmentation exchange features, raising segmentation accuracy on seven classes while detection stays neutral.","cross_cats":[],"primary_cat":"cs.CV","authors_text":"Ahmet \\.Inan\\c{c}, \\\"Ozg\\\"ur Erkent","submitted_at":"2026-04-14T16:00:53Z","abstract_excerpt":"Bird's-eye-view (BEV) representations are the dominant paradigm for 3D perception in autonomous driving, providing a unified spatial canvas where detection and segmentation features are geometrically registered to the same physical coordinate system. However, existing radar-camera fusion methods treat these tasks in isolation, missing the opportunity for cross-task feature sharing: object-level geometric cues from detection can sharpen segmentation, while dense road-layout context from segmentation can anchor detection. We propose \\textbf{CTAB} (Cross-Task Attention Bridge), a bidirectional mo"},"claims":{"count":4,"items":[{"kind":"strongest_claim","text":"CTAB is a bidirectional module that exchanges features between detection and segmentation branches via multi-scale deformable attention in shared BEV space.","source":"verdict.strongest_claim","status":"machine_extracted","claim_id":"C1","attestation":"unclaimed"},{"kind":"weakest_assumption","text":"That detection and segmentation features are sufficiently complementary and that attention-based transfer will not introduce noise or task interference that degrades either head.","source":"verdict.weakest_assumption","status":"machine_extracted","claim_id":"C2","attestation":"unclaimed"},{"kind":"one_line_summary","text":"CTAB exchanges features between detection and segmentation via multi-scale deformable attention in BEV space, yielding segmentation gains on 7 nuScenes classes at neutral detection cost.","source":"verdict.one_line_summary","status":"machine_extracted","claim_id":"C3","attestation":"unclaimed"},{"kind":"headline","text":"A bidirectional attention bridge in shared BEV space lets detection and segmentation exchange features, raising segmentation accuracy on seven classes while detection stays neutral.","source":"verdict.pith_extraction.headline","status":"machine_extracted","claim_id":"C4","attestation":"unclaimed"}],"snapshot_sha256":"5219f9b6892c128d779bc486b8cfa02a1295851a5b942c77ffeb484aba2397c1"},"source":{"id":"2604.12918","kind":"arxiv","version":2},"verdict":{"id":"e954fd38-0433-4a87-8332-bf91a7a20099","model_set":{"reader":"grok-4.3"},"created_at":"2026-05-10T15:09:50.716576Z","strongest_claim":"CTAB is a bidirectional module that exchanges features between detection and segmentation branches via multi-scale deformable attention in shared BEV space.","one_line_summary":"CTAB exchanges features between detection and segmentation via multi-scale deformable attention in BEV space, yielding segmentation gains on 7 nuScenes classes at neutral detection cost.","pipeline_version":"pith-pipeline@v0.9.0","weakest_assumption":"That detection and segmentation features are sufficiently complementary and that attention-based transfer will not introduce noise or task interference that degrades either head.","pith_extraction_headline":"A bidirectional attention bridge in shared BEV space lets detection and segmentation exchange features, raising segmentation accuracy on seven classes while detection stays neutral."},"integrity":{"clean":true,"summary":{"advisory":0,"critical":0,"by_detector":{},"informational":0},"endpoint":"/pith/2604.12918/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":"2604.12918","created_at":"2026-05-27T02:06:14.051110+00:00"},{"alias_kind":"arxiv_version","alias_value":"2604.12918v2","created_at":"2026-05-27T02:06:14.051110+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2604.12918","created_at":"2026-05-27T02:06:14.051110+00:00"},{"alias_kind":"pith_short_12","alias_value":"NMXAYK3NTGOH","created_at":"2026-05-27T02:06:14.051110+00:00"},{"alias_kind":"pith_short_16","alias_value":"NMXAYK3NTGOHEIQI","created_at":"2026-05-27T02:06:14.051110+00:00"},{"alias_kind":"pith_short_8","alias_value":"NMXAYK3N","created_at":"2026-05-27T02:06:14.051110+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":0,"internal_anchor_count":0,"sample":[]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/NMXAYK3NTGOHEIQIE6UPZMNQHO","json":"https://pith.science/pith/NMXAYK3NTGOHEIQIE6UPZMNQHO.json","graph_json":"https://pith.science/api/pith-number/NMXAYK3NTGOHEIQIE6UPZMNQHO/graph.json","events_json":"https://pith.science/api/pith-number/NMXAYK3NTGOHEIQIE6UPZMNQHO/events.json","paper":"https://pith.science/paper/NMXAYK3N"},"agent_actions":{"view_html":"https://pith.science/pith/NMXAYK3NTGOHEIQIE6UPZMNQHO","download_json":"https://pith.science/pith/NMXAYK3NTGOHEIQIE6UPZMNQHO.json","view_paper":"https://pith.science/paper/NMXAYK3N","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2604.12918&json=true","fetch_graph":"https://pith.science/api/pith-number/NMXAYK3NTGOHEIQIE6UPZMNQHO/graph.json","fetch_events":"https://pith.science/api/pith-number/NMXAYK3NTGOHEIQIE6UPZMNQHO/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/NMXAYK3NTGOHEIQIE6UPZMNQHO/action/timestamp_anchor","attest_storage":"https://pith.science/pith/NMXAYK3NTGOHEIQIE6UPZMNQHO/action/storage_attestation","attest_author":"https://pith.science/pith/NMXAYK3NTGOHEIQIE6UPZMNQHO/action/author_attestation","sign_citation":"https://pith.science/pith/NMXAYK3NTGOHEIQIE6UPZMNQHO/action/citation_signature","submit_replication":"https://pith.science/pith/NMXAYK3NTGOHEIQIE6UPZMNQHO/action/replication_record"}},"created_at":"2026-05-27T02:06:14.051110+00:00","updated_at":"2026-05-27T02:06:14.051110+00:00"}