{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2026:OJBTBE7GPCOVRGCXYCJ2ARLV3A","short_pith_number":"pith:OJBTBE7G","schema_version":"1.0","canonical_sha256":"72433093e6789d589857c093a04575d82f1ce0a0402ce3f018c7af3c21a82ed9","source":{"kind":"arxiv","id":"2604.03212","version":2},"attestation_state":"computed","paper":{"title":"ProtoFlow: Mitigating Forgetting in Class-Incremental Remote Sensing Segmentation via Low-Curvature Prototype Flow","license":"http://creativecommons.org/licenses/by-nc-sa/4.0/","headline":"ProtoFlow models class prototypes as low-curvature trajectories to reduce forgetting in continual remote sensing segmentation.","cross_cats":[],"primary_cat":"cs.CV","authors_text":"Amir H. Gandomi, Chuangqi Li, Dongxu Zhang, Guangxin Wu, Hao Zhang, Jianyuan Ni, Jiekai Wu, Pengbin Feng, Rong Fu, Shiyin Lin, Simon Fong, Yang Li, Zijian Zhang","submitted_at":"2026-04-03T17:40:58Z","abstract_excerpt":"Remote sensing segmentation in real deployment is inherently continual: new semantic categories emerge, and acquisition conditions shift across seasons, cities, and sensors. Despite recent progress, many incremental approaches still treat training steps as isolated updates, which leaves representation drift and forgetting insufficiently controlled. We present ProtoFlow, a time-aware prototype dynamics framework that models class prototypes as trajectories and learns their evolution with an explicit temporal vector field. By jointly enforcing low-curvature motion and inter-class separation, Pro"},"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":"2604.03212","kind":"arxiv","version":2},"metadata":{"license":"http://creativecommons.org/licenses/by-nc-sa/4.0/","primary_cat":"cs.CV","submitted_at":"2026-04-03T17:40:58Z","cross_cats_sorted":[],"title_canon_sha256":"54ea42daaad4a87e06007fa316cc16189c1a85464df14395fd78a3d56bec83fa","abstract_canon_sha256":"d05111ee1edd4b88258e178d8ec14a7b3af1f80c0d77030a06a0e61977350542"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-05-20T00:05:44.324543Z","signature_b64":"z4MjjY0WQbIFvHMUHa8186zDpNSwR9fP+CGjgIzMJQiXmk79GRa5iCcnXVQ6Zrk7tJ4zAdFF40vnq0G4Lo6+DQ==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"72433093e6789d589857c093a04575d82f1ce0a0402ce3f018c7af3c21a82ed9","last_reissued_at":"2026-05-20T00:05:44.323976Z","signature_status":"signed_v1","first_computed_at":"2026-05-20T00:05:44.323976Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"ProtoFlow: Mitigating Forgetting in Class-Incremental Remote Sensing Segmentation via Low-Curvature Prototype Flow","license":"http://creativecommons.org/licenses/by-nc-sa/4.0/","headline":"ProtoFlow models class prototypes as low-curvature trajectories to reduce forgetting in continual remote sensing segmentation.","cross_cats":[],"primary_cat":"cs.CV","authors_text":"Amir H. Gandomi, Chuangqi Li, Dongxu Zhang, Guangxin Wu, Hao Zhang, Jianyuan Ni, Jiekai Wu, Pengbin Feng, Rong Fu, Shiyin Lin, Simon Fong, Yang Li, Zijian Zhang","submitted_at":"2026-04-03T17:40:58Z","abstract_excerpt":"Remote sensing segmentation in real deployment is inherently continual: new semantic categories emerge, and acquisition conditions shift across seasons, cities, and sensors. Despite recent progress, many incremental approaches still treat training steps as isolated updates, which leaves representation drift and forgetting insufficiently controlled. We present ProtoFlow, a time-aware prototype dynamics framework that models class prototypes as trajectories and learns their evolution with an explicit temporal vector field. By jointly enforcing low-curvature motion and inter-class separation, Pro"},"claims":{"count":4,"items":[{"kind":"strongest_claim","text":"explicitly modeling temporal prototype evolution is a practical and interpretable strategy for robust continual remote sensing segmentation","source":"verdict.strongest_claim","status":"machine_extracted","claim_id":"C1","attestation":"unclaimed"},{"kind":"weakest_assumption","text":"That enforcing low-curvature motion and inter-class separation on learned prototype trajectories is sufficient to control representation drift and catastrophic forgetting without additional mechanisms such as replay buffers or architectural expansion.","source":"verdict.weakest_assumption","status":"machine_extracted","claim_id":"C2","attestation":"unclaimed"},{"kind":"one_line_summary","text":"ProtoFlow stabilizes class prototypes via low-curvature temporal flow to mitigate forgetting in class- and domain-incremental remote sensing segmentation.","source":"verdict.one_line_summary","status":"machine_extracted","claim_id":"C3","attestation":"unclaimed"},{"kind":"headline","text":"ProtoFlow models class prototypes as low-curvature trajectories to reduce forgetting in continual remote sensing segmentation.","source":"verdict.pith_extraction.headline","status":"machine_extracted","claim_id":"C4","attestation":"unclaimed"}],"snapshot_sha256":"0aed6b76c258314c937ba3e0713138dea9d8f972e8d2c779a9d88a757ac58c48"},"source":{"id":"2604.03212","kind":"arxiv","version":2},"verdict":{"id":"2c55d582-f9db-4f77-a63e-df96493947a3","model_set":{"reader":"grok-4.3"},"created_at":"2026-05-13T20:34:05.888684Z","strongest_claim":"explicitly modeling temporal prototype evolution is a practical and interpretable strategy for robust continual remote sensing segmentation","one_line_summary":"ProtoFlow stabilizes class prototypes via low-curvature temporal flow to mitigate forgetting in class- and domain-incremental remote sensing segmentation.","pipeline_version":"pith-pipeline@v0.9.0","weakest_assumption":"That enforcing low-curvature motion and inter-class separation on learned prototype trajectories is sufficient to control representation drift and catastrophic forgetting without additional mechanisms such as replay buffers or architectural expansion.","pith_extraction_headline":"ProtoFlow models class prototypes as low-curvature trajectories to reduce forgetting in continual remote sensing segmentation."},"integrity":{"clean":true,"summary":{"advisory":0,"critical":0,"by_detector":{},"informational":0},"endpoint":"/pith/2604.03212/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":2,"snapshot_sha256":"6e27e40c68c44e1dbb62b5b241648134300f15e55092f02a4233223a63908f71"},"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":"2604.03212","created_at":"2026-05-20T00:05:44.324070+00:00"},{"alias_kind":"arxiv_version","alias_value":"2604.03212v2","created_at":"2026-05-20T00:05:44.324070+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2604.03212","created_at":"2026-05-20T00:05:44.324070+00:00"},{"alias_kind":"pith_short_12","alias_value":"OJBTBE7GPCOV","created_at":"2026-05-20T00:05:44.324070+00:00"},{"alias_kind":"pith_short_16","alias_value":"OJBTBE7GPCOVRGCX","created_at":"2026-05-20T00:05:44.324070+00:00"},{"alias_kind":"pith_short_8","alias_value":"OJBTBE7G","created_at":"2026-05-20T00:05:44.324070+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":1,"internal_anchor_count":1,"sample":[{"citing_arxiv_id":"2605.01732","citing_title":"EGAD: Entropy-Guided Adaptive Distillation for Token-Level Knowledge Transfer","ref_index":39,"is_internal_anchor":true}]},"formal_canon":{"evidence_count":2,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/OJBTBE7GPCOVRGCXYCJ2ARLV3A","json":"https://pith.science/pith/OJBTBE7GPCOVRGCXYCJ2ARLV3A.json","graph_json":"https://pith.science/api/pith-number/OJBTBE7GPCOVRGCXYCJ2ARLV3A/graph.json","events_json":"https://pith.science/api/pith-number/OJBTBE7GPCOVRGCXYCJ2ARLV3A/events.json","paper":"https://pith.science/paper/OJBTBE7G"},"agent_actions":{"view_html":"https://pith.science/pith/OJBTBE7GPCOVRGCXYCJ2ARLV3A","download_json":"https://pith.science/pith/OJBTBE7GPCOVRGCXYCJ2ARLV3A.json","view_paper":"https://pith.science/paper/OJBTBE7G","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2604.03212&json=true","fetch_graph":"https://pith.science/api/pith-number/OJBTBE7GPCOVRGCXYCJ2ARLV3A/graph.json","fetch_events":"https://pith.science/api/pith-number/OJBTBE7GPCOVRGCXYCJ2ARLV3A/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/OJBTBE7GPCOVRGCXYCJ2ARLV3A/action/timestamp_anchor","attest_storage":"https://pith.science/pith/OJBTBE7GPCOVRGCXYCJ2ARLV3A/action/storage_attestation","attest_author":"https://pith.science/pith/OJBTBE7GPCOVRGCXYCJ2ARLV3A/action/author_attestation","sign_citation":"https://pith.science/pith/OJBTBE7GPCOVRGCXYCJ2ARLV3A/action/citation_signature","submit_replication":"https://pith.science/pith/OJBTBE7GPCOVRGCXYCJ2ARLV3A/action/replication_record"}},"created_at":"2026-05-20T00:05:44.324070+00:00","updated_at":"2026-05-20T00:05:44.324070+00:00"}