{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2025:WOPPOUGGU5W3X7KEWSDRV7YCQF","short_pith_number":"pith:WOPPOUGG","schema_version":"1.0","canonical_sha256":"b39ef750c6a76dbbfd44b4871aff02815a5fe038f3d2c36f8f10e0d9ec007ce7","source":{"kind":"arxiv","id":"2512.12307","version":5},"attestation_state":"computed","paper":{"title":"MRD: Using Physically Based Differentiable Rendering to Probe Vision Models for 3D Scene Understanding","license":"http://creativecommons.org/licenses/by-nc-sa/4.0/","headline":"","cross_cats":["cs.GR"],"primary_cat":"cs.CV","authors_text":"Benjamin Beilharz, Thomas S. A. Wallis","submitted_at":"2025-12-13T12:26:57Z","abstract_excerpt":"While deep learning methods have achieved impressive success in many vision benchmarks, it remains difficult to understand and explain the representations and decisions of these models. Though vision models are typically trained on 2D inputs, they are often assumed to develop an implicit representation of the underlying 3D scene (for example, showing tolerance to partial occlusion, or the ability to reason about relative depth). Here, we introduce MRD (metamers rendered differentiably), an approach that uses physically based differentiable rendering to probe vision models' implicit understandi"},"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":"2512.12307","kind":"arxiv","version":5},"metadata":{"license":"http://creativecommons.org/licenses/by-nc-sa/4.0/","primary_cat":"cs.CV","submitted_at":"2025-12-13T12:26:57Z","cross_cats_sorted":["cs.GR"],"title_canon_sha256":"be59b4e9e9375c8734ece525d438c93a13313ad3de53703227cb9468999e5dbd","abstract_canon_sha256":"4b825ec5bc64202ca4177257427cdf787946bb39895984d79d6b83dbc5f0dc30"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"b39ef750c6a76dbbfd44b4871aff02815a5fe038f3d2c36f8f10e0d9ec007ce7","last_reissued_at":"2026-07-31T01:33:01.473396Z","signature_status":"unsigned_v0","first_computed_at":"2026-07-31T01:33:01.473396Z"},"graph_snapshot":{"paper":{"title":"MRD: Using Physically Based Differentiable Rendering to Probe Vision Models for 3D Scene Understanding","license":"http://creativecommons.org/licenses/by-nc-sa/4.0/","headline":"","cross_cats":["cs.GR"],"primary_cat":"cs.CV","authors_text":"Benjamin Beilharz, Thomas S. A. Wallis","submitted_at":"2025-12-13T12:26:57Z","abstract_excerpt":"While deep learning methods have achieved impressive success in many vision benchmarks, it remains difficult to understand and explain the representations and decisions of these models. Though vision models are typically trained on 2D inputs, they are often assumed to develop an implicit representation of the underlying 3D scene (for example, showing tolerance to partial occlusion, or the ability to reason about relative depth). Here, we introduce MRD (metamers rendered differentiably), an approach that uses physically based differentiable rendering to probe vision models' implicit understandi"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2512.12307","kind":"arxiv","version":5},"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/2512.12307/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":"2512.12307","created_at":"2026-07-31T01:33:01.477985+00:00"},{"alias_kind":"arxiv_version","alias_value":"2512.12307v5","created_at":"2026-07-31T01:33:01.477985+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2512.12307","created_at":"2026-07-31T01:33:01.477985+00:00"},{"alias_kind":"pith_short_12","alias_value":"WOPPOUGGU5W3","created_at":"2026-07-31T01:33:01.477985+00:00"},{"alias_kind":"pith_short_16","alias_value":"WOPPOUGGU5W3X7KE","created_at":"2026-07-31T01:33:01.477985+00:00"},{"alias_kind":"pith_short_8","alias_value":"WOPPOUGG","created_at":"2026-07-31T01:33:01.477985+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":1,"internal_anchor_count":1,"sample":[{"citing_arxiv_id":"2605.20549","citing_title":"MAPS: A Synthetic Dataset for Probing Vision Models in a Controlled 3D Scene Space","ref_index":8,"is_internal_anchor":true}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/WOPPOUGGU5W3X7KEWSDRV7YCQF","json":"https://pith.science/pith/WOPPOUGGU5W3X7KEWSDRV7YCQF.json","graph_json":"https://pith.science/api/pith-number/WOPPOUGGU5W3X7KEWSDRV7YCQF/graph.json","events_json":"https://pith.science/api/pith-number/WOPPOUGGU5W3X7KEWSDRV7YCQF/events.json","paper":"https://pith.science/paper/WOPPOUGG"},"agent_actions":{"view_html":"https://pith.science/pith/WOPPOUGGU5W3X7KEWSDRV7YCQF","download_json":"https://pith.science/pith/WOPPOUGGU5W3X7KEWSDRV7YCQF.json","view_paper":"https://pith.science/paper/WOPPOUGG","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2512.12307&json=true","fetch_graph":"https://pith.science/api/pith-number/WOPPOUGGU5W3X7KEWSDRV7YCQF/graph.json","fetch_events":"https://pith.science/api/pith-number/WOPPOUGGU5W3X7KEWSDRV7YCQF/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/WOPPOUGGU5W3X7KEWSDRV7YCQF/action/timestamp_anchor","attest_storage":"https://pith.science/pith/WOPPOUGGU5W3X7KEWSDRV7YCQF/action/storage_attestation","attest_author":"https://pith.science/pith/WOPPOUGGU5W3X7KEWSDRV7YCQF/action/author_attestation","sign_citation":"https://pith.science/pith/WOPPOUGGU5W3X7KEWSDRV7YCQF/action/citation_signature","submit_replication":"https://pith.science/pith/WOPPOUGGU5W3X7KEWSDRV7YCQF/action/replication_record"}},"created_at":"2026-07-31T01:33:01.477985+00:00","updated_at":"2026-07-31T01:33:01.477985+00:00"}