{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2025:W5YI2V23IIQOSNM4N6VR2VHRIP","short_pith_number":"pith:W5YI2V23","schema_version":"1.0","canonical_sha256":"b7708d575b4220e9359c6fab1d54f143eecd1e858758c1d1f08abc699c363b4e","source":{"kind":"arxiv","id":"2508.07402","version":2},"attestation_state":"computed","paper":{"title":"ForensicsSAM: Toward Robust and Unified Image Forgery Detection and Localization Resisting to Adversarial Attack","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":[],"primary_cat":"cs.CV","authors_text":"Alex C. Kot, Anwei Luo, Chenqi Kong, Jiwu Huang, Rongxuan Peng, Shunquan Tan","submitted_at":"2025-08-10T16:03:44Z","abstract_excerpt":"Parameter-efficient fine-tuning (PEFT) has emerged as a popular strategy for adapting large vision foundation models, such as the Segment Anything Model (SAM) and LLaVA, to downstream tasks like image forgery detection and localization (IFDL). However, existing PEFT-based approaches overlook their vulnerability to adversarial attacks. In this paper, we show that highly transferable adversarial images can be crafted solely via the upstream model, without accessing the downstream model or training data, significantly degrading the IFDL performance. To address this, we propose ForensicsSAM, a uni"},"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":"2508.07402","kind":"arxiv","version":2},"metadata":{"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.CV","submitted_at":"2025-08-10T16:03:44Z","cross_cats_sorted":[],"title_canon_sha256":"13700e4dc7a793cf519da1e46bb2950b7b124e20119d18373390dee59389ea21","abstract_canon_sha256":"56b53ff974709e02c8d99ad7a1e95be2b323b089d84a40526f08d8c33d65969a"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T11:54:50.699042Z","signature_b64":"YXy7aH+2rQmCUiwEZHx0XCfLrtcJ5ymcVVQstG7x/yq4BmCs3HjIBogCD3bxddguTLAy2buWKFOBQnkLD5zEBg==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"b7708d575b4220e9359c6fab1d54f143eecd1e858758c1d1f08abc699c363b4e","last_reissued_at":"2026-07-05T11:54:50.698563Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T11:54:50.698563Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"ForensicsSAM: Toward Robust and Unified Image Forgery Detection and Localization Resisting to Adversarial Attack","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":[],"primary_cat":"cs.CV","authors_text":"Alex C. Kot, Anwei Luo, Chenqi Kong, Jiwu Huang, Rongxuan Peng, Shunquan Tan","submitted_at":"2025-08-10T16:03:44Z","abstract_excerpt":"Parameter-efficient fine-tuning (PEFT) has emerged as a popular strategy for adapting large vision foundation models, such as the Segment Anything Model (SAM) and LLaVA, to downstream tasks like image forgery detection and localization (IFDL). However, existing PEFT-based approaches overlook their vulnerability to adversarial attacks. In this paper, we show that highly transferable adversarial images can be crafted solely via the upstream model, without accessing the downstream model or training data, significantly degrading the IFDL performance. To address this, we propose ForensicsSAM, a uni"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2508.07402","kind":"arxiv","version":2},"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/2508.07402/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":"2508.07402","created_at":"2026-07-05T11:54:50.698621+00:00"},{"alias_kind":"arxiv_version","alias_value":"2508.07402v2","created_at":"2026-07-05T11:54:50.698621+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2508.07402","created_at":"2026-07-05T11:54:50.698621+00:00"},{"alias_kind":"pith_short_12","alias_value":"W5YI2V23IIQO","created_at":"2026-07-05T11:54:50.698621+00:00"},{"alias_kind":"pith_short_16","alias_value":"W5YI2V23IIQOSNM4","created_at":"2026-07-05T11:54:50.698621+00:00"},{"alias_kind":"pith_short_8","alias_value":"W5YI2V23","created_at":"2026-07-05T11:54:50.698621+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":2,"internal_anchor_count":0,"sample":[{"citing_arxiv_id":"2605.28609","citing_title":"JECA^2: Judgment-Explanation Consistent Adversarial Attack against Forensic Vision-Language Models","ref_index":24,"is_internal_anchor":false},{"citing_arxiv_id":"2604.09096","citing_title":"Off-the-shelf Vision Models Benefit Image Manipulation Localization","ref_index":38,"is_internal_anchor":false}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/W5YI2V23IIQOSNM4N6VR2VHRIP","json":"https://pith.science/pith/W5YI2V23IIQOSNM4N6VR2VHRIP.json","graph_json":"https://pith.science/api/pith-number/W5YI2V23IIQOSNM4N6VR2VHRIP/graph.json","events_json":"https://pith.science/api/pith-number/W5YI2V23IIQOSNM4N6VR2VHRIP/events.json","paper":"https://pith.science/paper/W5YI2V23"},"agent_actions":{"view_html":"https://pith.science/pith/W5YI2V23IIQOSNM4N6VR2VHRIP","download_json":"https://pith.science/pith/W5YI2V23IIQOSNM4N6VR2VHRIP.json","view_paper":"https://pith.science/paper/W5YI2V23","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2508.07402&json=true","fetch_graph":"https://pith.science/api/pith-number/W5YI2V23IIQOSNM4N6VR2VHRIP/graph.json","fetch_events":"https://pith.science/api/pith-number/W5YI2V23IIQOSNM4N6VR2VHRIP/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/W5YI2V23IIQOSNM4N6VR2VHRIP/action/timestamp_anchor","attest_storage":"https://pith.science/pith/W5YI2V23IIQOSNM4N6VR2VHRIP/action/storage_attestation","attest_author":"https://pith.science/pith/W5YI2V23IIQOSNM4N6VR2VHRIP/action/author_attestation","sign_citation":"https://pith.science/pith/W5YI2V23IIQOSNM4N6VR2VHRIP/action/citation_signature","submit_replication":"https://pith.science/pith/W5YI2V23IIQOSNM4N6VR2VHRIP/action/replication_record"}},"created_at":"2026-07-05T11:54:50.698621+00:00","updated_at":"2026-07-05T11:54:50.698621+00:00"}