{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2026:LHFLFPOVLOCCYTTPKQLAKQT2U2","short_pith_number":"pith:LHFLFPOV","schema_version":"1.0","canonical_sha256":"59cab2bdd55b842c4e6f541605427aa681fe387743014d38214f25662c9f923c","source":{"kind":"arxiv","id":"2605.07379","version":2},"attestation_state":"computed","paper":{"title":"RELO: Reinforcement Learning to Localize for Visual Object Tracking","license":"http://creativecommons.org/licenses/by/4.0/","headline":"RELO learns a reinforcement learning policy to localize targets in visual tracking by maximizing IoU and AUC rewards instead of relying on handcrafted spatial priors.","cross_cats":["cs.AI"],"primary_cat":"cs.CV","authors_text":"Chuanyu Sun, Dong Wang, Houwen Peng, Huchuan Lu, Jiao Xu, Kede Ma, Xin Chen","submitted_at":"2026-05-08T07:34:29Z","abstract_excerpt":"Conventional visual object trackers localize targets using handcrafted spatial priors, often in the form of heatmaps. Such priors provide only surrogate supervision and are poorly aligned with tracking optimization and evaluation metrics, such as intersection over union (IoU) and area under the success curve (AUC). Here, we introduce RELO, a REinforcement-learning-to-LOcalize method for visual object tracking that formulates target localization as a Markov decision process. Specifically, RELO replaces handcrafted spatial priors with a localization policy learned over spatial positions via rein"},"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":"2605.07379","kind":"arxiv","version":2},"metadata":{"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.CV","submitted_at":"2026-05-08T07:34:29Z","cross_cats_sorted":["cs.AI"],"title_canon_sha256":"36e429072431a8e3c3669e4e2dec412c54b56d855cbf16207eb3c8c59ed4207c","abstract_canon_sha256":"e242862be18eb9f789b5eb46325ddf33737f82ed5a93c88d1653deb0033fdb0d"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-05-20T01:05:15.748670Z","signature_b64":"Pa9XR2tcpm7b+wBDgzS1N+Hqwmd+YOFvIrdAOvUI9AAQxFm0hUUxiJ9+x6GMrabdcVirac0sHtNDT9VW4th8Aw==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"59cab2bdd55b842c4e6f541605427aa681fe387743014d38214f25662c9f923c","last_reissued_at":"2026-05-20T01:05:15.747775Z","signature_status":"signed_v1","first_computed_at":"2026-05-20T01:05:15.747775Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"RELO: Reinforcement Learning to Localize for Visual Object Tracking","license":"http://creativecommons.org/licenses/by/4.0/","headline":"RELO learns a reinforcement learning policy to localize targets in visual tracking by maximizing IoU and AUC rewards instead of relying on handcrafted spatial priors.","cross_cats":["cs.AI"],"primary_cat":"cs.CV","authors_text":"Chuanyu Sun, Dong Wang, Houwen Peng, Huchuan Lu, Jiao Xu, Kede Ma, Xin Chen","submitted_at":"2026-05-08T07:34:29Z","abstract_excerpt":"Conventional visual object trackers localize targets using handcrafted spatial priors, often in the form of heatmaps. Such priors provide only surrogate supervision and are poorly aligned with tracking optimization and evaluation metrics, such as intersection over union (IoU) and area under the success curve (AUC). Here, we introduce RELO, a REinforcement-learning-to-LOcalize method for visual object tracking that formulates target localization as a Markov decision process. Specifically, RELO replaces handcrafted spatial priors with a localization policy learned over spatial positions via rein"},"claims":{"count":4,"items":[{"kind":"strongest_claim","text":"RELO replaces handcrafted spatial priors with a localization policy learned over spatial positions via reinforcement learning, with rewards combining frame-level IoU and sequence-level AUC, attaining 57.5% AUC on LaSOText without template updates.","source":"verdict.strongest_claim","status":"machine_extracted","claim_id":"C1","attestation":"unclaimed"},{"kind":"weakest_assumption","text":"That a policy trained with combined IoU and AUC rewards will generalize across diverse tracking scenarios and benchmarks without requiring template updates or extensive hyperparameter tuning.","source":"verdict.weakest_assumption","status":"machine_extracted","claim_id":"C2","attestation":"unclaimed"},{"kind":"one_line_summary","text":"RELO replaces handcrafted spatial priors with a reinforcement learning policy for target localization in visual tracking and reports 57.5% AUC on LaSOText without template updates.","source":"verdict.one_line_summary","status":"machine_extracted","claim_id":"C3","attestation":"unclaimed"},{"kind":"headline","text":"RELO learns a reinforcement learning policy to localize targets in visual tracking by maximizing IoU and AUC rewards instead of relying on handcrafted spatial priors.","source":"verdict.pith_extraction.headline","status":"machine_extracted","claim_id":"C4","attestation":"unclaimed"}],"snapshot_sha256":"d5aea890770e1be8fabd4c96e6553a5f9892b4c25eff3d91abab0f37250d8d3d"},"source":{"id":"2605.07379","kind":"arxiv","version":2},"verdict":{"id":"53cf90d1-79f3-4b5e-9d93-0ede0db77676","model_set":{"reader":"grok-4.3"},"created_at":"2026-05-11T02:17:48.249265Z","strongest_claim":"RELO replaces handcrafted spatial priors with a localization policy learned over spatial positions via reinforcement learning, with rewards combining frame-level IoU and sequence-level AUC, attaining 57.5% AUC on LaSOText without template updates.","one_line_summary":"RELO replaces handcrafted spatial priors with a reinforcement learning policy for target localization in visual tracking and reports 57.5% AUC on LaSOText without template updates.","pipeline_version":"pith-pipeline@v0.9.0","weakest_assumption":"That a policy trained with combined IoU and AUC rewards will generalize across diverse tracking scenarios and benchmarks without requiring template updates or extensive hyperparameter tuning.","pith_extraction_headline":"RELO learns a reinforcement learning policy to localize targets in visual tracking by maximizing IoU and AUC rewards instead of relying on handcrafted spatial priors."},"integrity":{"clean":true,"summary":{"advisory":0,"critical":0,"by_detector":{},"informational":0},"endpoint":"/pith/2605.07379/integrity.json","findings":[],"available":true,"detectors_run":[{"name":"doi_title_agreement","ran_at":"2026-05-19T16:31:19.350282Z","status":"completed","version":"1.0.0","findings_count":0},{"name":"doi_compliance","ran_at":"2026-05-19T11:52:52.117022Z","status":"completed","version":"1.0.0","findings_count":0}],"snapshot_sha256":"c02416f2298f5ee1f95cf7e2a8cd9f915c94e7cab6f3fde04d26593b9ae8596a"},"references":{"count":0,"sample":[],"resolved_work":0,"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57","internal_anchors":0},"formal_canon":{"evidence_count":2,"snapshot_sha256":"241098666ded0c70c9c0a8b8226d6e30ef4f7448e596834a6eca88331ed960c7"},"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":"2605.07379","created_at":"2026-05-20T01:05:15.747882+00:00"},{"alias_kind":"arxiv_version","alias_value":"2605.07379v2","created_at":"2026-05-20T01:05:15.747882+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2605.07379","created_at":"2026-05-20T01:05:15.747882+00:00"},{"alias_kind":"pith_short_12","alias_value":"LHFLFPOVLOCC","created_at":"2026-05-20T01:05:15.747882+00:00"},{"alias_kind":"pith_short_16","alias_value":"LHFLFPOVLOCCYTTP","created_at":"2026-05-20T01:05:15.747882+00:00"},{"alias_kind":"pith_short_8","alias_value":"LHFLFPOV","created_at":"2026-05-20T01:05:15.747882+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/LHFLFPOVLOCCYTTPKQLAKQT2U2","json":"https://pith.science/pith/LHFLFPOVLOCCYTTPKQLAKQT2U2.json","graph_json":"https://pith.science/api/pith-number/LHFLFPOVLOCCYTTPKQLAKQT2U2/graph.json","events_json":"https://pith.science/api/pith-number/LHFLFPOVLOCCYTTPKQLAKQT2U2/events.json","paper":"https://pith.science/paper/LHFLFPOV"},"agent_actions":{"view_html":"https://pith.science/pith/LHFLFPOVLOCCYTTPKQLAKQT2U2","download_json":"https://pith.science/pith/LHFLFPOVLOCCYTTPKQLAKQT2U2.json","view_paper":"https://pith.science/paper/LHFLFPOV","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2605.07379&json=true","fetch_graph":"https://pith.science/api/pith-number/LHFLFPOVLOCCYTTPKQLAKQT2U2/graph.json","fetch_events":"https://pith.science/api/pith-number/LHFLFPOVLOCCYTTPKQLAKQT2U2/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/LHFLFPOVLOCCYTTPKQLAKQT2U2/action/timestamp_anchor","attest_storage":"https://pith.science/pith/LHFLFPOVLOCCYTTPKQLAKQT2U2/action/storage_attestation","attest_author":"https://pith.science/pith/LHFLFPOVLOCCYTTPKQLAKQT2U2/action/author_attestation","sign_citation":"https://pith.science/pith/LHFLFPOVLOCCYTTPKQLAKQT2U2/action/citation_signature","submit_replication":"https://pith.science/pith/LHFLFPOVLOCCYTTPKQLAKQT2U2/action/replication_record"}},"created_at":"2026-05-20T01:05:15.747882+00:00","updated_at":"2026-05-20T01:05:15.747882+00:00"}