{"bundle_type":"pith_open_graph_bundle","bundle_version":"1.0","pith_number":"pith:2026:NDFU7LT7BTHDNNGYV54MVJJVSD","short_pith_number":"pith:NDFU7LT7","canonical_record":{"source":{"id":"2605.13396","kind":"arxiv","version":1},"metadata":{"license":"http://creativecommons.org/licenses/by-nc-sa/4.0/","primary_cat":"cs.CV","submitted_at":"2026-05-13T11:53:19Z","cross_cats_sorted":[],"title_canon_sha256":"f5d0e16b63a6af67a943f55eefeeb4a9ba81386f2424d207404c05bfb85ba27d","abstract_canon_sha256":"9e4cdcaee995b076a8ff0dec95a8c56b9d0790d1ac75b6b49333f09945cb0f66"},"schema_version":"1.0"},"canonical_sha256":"68cb4fae7f0cce36b4d8af78caa53590e46729fd458e07a8b3a715fb575cd129","source":{"kind":"arxiv","id":"2605.13396","version":1},"source_aliases":[{"alias_kind":"arxiv","alias_value":"2605.13396","created_at":"2026-05-18T02:44:47Z"},{"alias_kind":"arxiv_version","alias_value":"2605.13396v1","created_at":"2026-05-18T02:44:47Z"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2605.13396","created_at":"2026-05-18T02:44:47Z"},{"alias_kind":"pith_short_12","alias_value":"NDFU7LT7BTHD","created_at":"2026-05-18T12:33:37Z"},{"alias_kind":"pith_short_16","alias_value":"NDFU7LT7BTHDNNGY","created_at":"2026-05-18T12:33:37Z"},{"alias_kind":"pith_short_8","alias_value":"NDFU7LT7","created_at":"2026-05-18T12:33:37Z"}],"events":[{"event_type":"record_created","subject_pith_number":"pith:2026:NDFU7LT7BTHDNNGYV54MVJJVSD","target":"record","payload":{"canonical_record":{"source":{"id":"2605.13396","kind":"arxiv","version":1},"metadata":{"license":"http://creativecommons.org/licenses/by-nc-sa/4.0/","primary_cat":"cs.CV","submitted_at":"2026-05-13T11:53:19Z","cross_cats_sorted":[],"title_canon_sha256":"f5d0e16b63a6af67a943f55eefeeb4a9ba81386f2424d207404c05bfb85ba27d","abstract_canon_sha256":"9e4cdcaee995b076a8ff0dec95a8c56b9d0790d1ac75b6b49333f09945cb0f66"},"schema_version":"1.0"},"canonical_sha256":"68cb4fae7f0cce36b4d8af78caa53590e46729fd458e07a8b3a715fb575cd129","receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-05-18T02:44:47.652351Z","signature_b64":"kd7vKcYtbrCc86VeUmMHpGqU0tC/I7KgfuZv/UCOrqJSsS29a6b5C0fHz1APWsI+kt/tDxm9zlh3osiNSChiDQ==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"68cb4fae7f0cce36b4d8af78caa53590e46729fd458e07a8b3a715fb575cd129","last_reissued_at":"2026-05-18T02:44:47.651886Z","signature_status":"signed_v1","first_computed_at":"2026-05-18T02:44:47.651886Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"source_kind":"arxiv","source_id":"2605.13396","source_version":1,"attestation_state":"computed"},"signer":{"signer_id":"pith.science","signer_type":"pith_registry","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"created_at":"2026-05-18T02:44:47Z","supersedes":[],"prev_event":null,"signature":{"signature_status":"signed_v1","algorithm":"ed25519","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signature_b64":"wX3jQ61MsWNDhZf/qJhciUGV7ecUnjh78gxbz9jG1j5Ld12d4BxafzskJbvEZUnTWGtItIrjO2nMCM3Ewmt9Bw==","signed_message":"open_graph_event_sha256_bytes","signed_at":"2026-06-06T19:20:11.882226Z"},"content_sha256":"c77a9dea1207fd7e4cad7d596d62159ebe1d59d7fa43b5c38650b90e3c9a9c03","schema_version":"1.0","event_id":"sha256:c77a9dea1207fd7e4cad7d596d62159ebe1d59d7fa43b5c38650b90e3c9a9c03"},{"event_type":"graph_snapshot","subject_pith_number":"pith:2026:NDFU7LT7BTHDNNGYV54MVJJVSD","target":"graph","payload":{"graph_snapshot":{"paper":{"title":"PreFIQs: Face Image Quality Is What Survives Pruning","license":"http://creativecommons.org/licenses/by-nc-sa/4.0/","headline":"Face image quality equals the embedding shift that occurs when a face recognition model is pruned.","cross_cats":[],"primary_cat":"cs.CV","authors_text":"Andrea Atzori, Fadi Boutros, Guray Ozgur, Jan Niklas Kolf, Naser Damer, Vitomir \\v{S}truc, \\v{Z}iga Babnik","submitted_at":"2026-05-13T11:53:19Z","abstract_excerpt":"Face Image Quality Assessment (FIQA) evaluates the utility of a face image for automated face recognition (FR) systems. In this work, we propose PreFIQs, an unsupervised and training-free FIQA framework grounded in the Pruning Identified Exemplar (PIE) hypothesis. We hypothesize that low-utility face images rely disproportionately on fragile network parameters, resulting in larger geometric displacement of their embeddings under model sparsification. Accordingly, PreFIQs quantifies image utility as the Euclidean distance between L2-normalized embeddings extracted from a pre-trained FR model an"},"claims":{"count":4,"items":[{"kind":"strongest_claim","text":"PreFIQs quantifies image utility as the Euclidean distance between L2-normalized embeddings extracted from a pre-trained FR model and its pruned counterpart... achieves competitive or superior performance compared to state-of-the-art FIQA methods, including establishing new state-of-the-art results on several benchmarks, without any training or supervision.","source":"verdict.strongest_claim","status":"machine_extracted","claim_id":"C1","attestation":"unclaimed"},{"kind":"weakest_assumption","text":"We hypothesize that low-utility face images rely disproportionately on fragile network parameters, resulting in larger geometric displacement of their embeddings under model sparsification.","source":"verdict.weakest_assumption","status":"machine_extracted","claim_id":"C2","attestation":"unclaimed"},{"kind":"one_line_summary","text":"Face image quality is quantified as the Euclidean distance between embeddings from a pre-trained face recognition model and its pruned version, achieving competitive or superior results without training or supervision.","source":"verdict.one_line_summary","status":"machine_extracted","claim_id":"C3","attestation":"unclaimed"},{"kind":"headline","text":"Face image quality equals the embedding shift that occurs when a face recognition model is pruned.","source":"verdict.pith_extraction.headline","status":"machine_extracted","claim_id":"C4","attestation":"unclaimed"}],"snapshot_sha256":"a220ca0bccfa0dd97d5b59f9bf5d9572e5f76f30a8bae47ca9e4d29ba7ea15e2"},"source":{"id":"2605.13396","kind":"arxiv","version":1},"verdict":{"id":"92c26ae3-2e6d-4d5d-9cf3-fee4be08eb71","model_set":{"reader":"grok-4.3"},"created_at":"2026-05-14T20:46:31.142595Z","strongest_claim":"PreFIQs quantifies image utility as the Euclidean distance between L2-normalized embeddings extracted from a pre-trained FR model and its pruned counterpart... achieves competitive or superior performance compared to state-of-the-art FIQA methods, including establishing new state-of-the-art results on several benchmarks, without any training or supervision.","one_line_summary":"Face image quality is quantified as the Euclidean distance between embeddings from a pre-trained face recognition model and its pruned version, achieving competitive or superior results without training or supervision.","pipeline_version":"pith-pipeline@v0.9.0","weakest_assumption":"We hypothesize that low-utility face images rely disproportionately on fragile network parameters, resulting in larger geometric displacement of their embeddings under model sparsification.","pith_extraction_headline":"Face image quality equals the embedding shift that occurs when a face recognition model is pruned."},"references":{"count":56,"sample":[{"doi":"","year":2025,"title":"Deep network pruning: A comparative study on cnns in face recognition","work_id":"fc28e841-f615-4c77-8542-37f07d28c01b","ref_index":1,"cited_arxiv_id":"","is_internal_anchor":false},{"doi":"","year":2025,"title":"Vit- fiqa: Assessing face image quality using vision trans- formers","work_id":"0869dcd4-8792-43a0-8981-8f6a211d5956","ref_index":2,"cited_arxiv_id":"","is_internal_anchor":false},{"doi":"","year":2022,"title":"Faceqan: Face image quality assessment through adversarial noise exploration","work_id":"0417d38a-ab6b-4015-ac1c-4d322b5b1bc0","ref_index":3,"cited_arxiv_id":"","is_internal_anchor":false},{"doi":"","year":2023,"title":"Diffiqa: Face image quality assessment using denoising dif- fusion probabilistic models","work_id":"7c65940e-2dfb-4c9d-bc32-96a2ab943867","ref_index":4,"cited_arxiv_id":"","is_internal_anchor":false},{"doi":"","year":2024,"title":"eDifFIQA: Towards Efficient Face Image Quality Assessment based on Denoising Diffusion Probabilistic Models","work_id":"e5c985eb-0296-4e03-9097-f1b2a0633046","ref_index":5,"cited_arxiv_id":"","is_internal_anchor":false}],"resolved_work":56,"snapshot_sha256":"bfa0e7d8ea8a01502e9a30f3290f9909ddf7e56589d667e2f5d2249cd5199bf2","internal_anchors":2},"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"},"verdict_id":"92c26ae3-2e6d-4d5d-9cf3-fee4be08eb71"},"signer":{"signer_id":"pith.science","signer_type":"pith_registry","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"created_at":"2026-05-18T02:44:47Z","supersedes":[],"prev_event":null,"signature":{"signature_status":"signed_v1","algorithm":"ed25519","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signature_b64":"2Bf/gvAlAR1VCK53eSkaHMXnaecC5974rDqSn817sYobv2KyC/QXsrtGKhCHU7HGLyvwhT9/Ko4v2/O3Bi5KDA==","signed_message":"open_graph_event_sha256_bytes","signed_at":"2026-06-06T19:20:11.882737Z"},"content_sha256":"bd7858d65bedb46387d0af5db8a6603c84596ab1d0d9a580ab867c7ffe6d4773","schema_version":"1.0","event_id":"sha256:bd7858d65bedb46387d0af5db8a6603c84596ab1d0d9a580ab867c7ffe6d4773"}],"timestamp_proofs":[],"mirror_hints":[{"mirror_type":"https","name":"Pith Resolver","base_url":"https://pith.science","bundle_url":"https://pith.science/pith/NDFU7LT7BTHDNNGYV54MVJJVSD/bundle.json","state_url":"https://pith.science/pith/NDFU7LT7BTHDNNGYV54MVJJVSD/state.json","well_known_bundle_url":"https://pith.science/.well-known/pith/NDFU7LT7BTHDNNGYV54MVJJVSD/bundle.json","status":"primary"}],"public_keys":[{"key_id":"pith-v1-2026-05","algorithm":"ed25519","format":"raw","public_key_b64":"stVStoiQhXFxp4s2pdzPNoqVNBMojDU/fJ2db5S3CbM=","public_key_hex":"b2d552b68890857171a78b36a5dccf368a953413288c353f7c9d9d6f94b709b3","fingerprint_sha256_b32_first128bits":"RVFV5Z2OI2J3ZUO7ERDEBCYNKS","fingerprint_sha256_hex":"8d4b5ee74e4693bcd1df2446408b0d54","rotates_at":null,"url":"https://pith.science/pith-signing-key.json","notes":"Pith uses this Ed25519 key to sign canonical record SHA-256 digests. Verify with: ed25519_verify(public_key, message=canonical_sha256_bytes, signature=base64decode(signature_b64))."}],"merge_version":"pith-open-graph-merge-v1","built_at":"2026-06-06T19:20:11Z","links":{"resolver":"https://pith.science/pith/NDFU7LT7BTHDNNGYV54MVJJVSD","bundle":"https://pith.science/pith/NDFU7LT7BTHDNNGYV54MVJJVSD/bundle.json","state":"https://pith.science/pith/NDFU7LT7BTHDNNGYV54MVJJVSD/state.json","well_known_bundle":"https://pith.science/.well-known/pith/NDFU7LT7BTHDNNGYV54MVJJVSD/bundle.json"},"state":{"state_type":"pith_open_graph_state","state_version":"1.0","pith_number":"pith:2026:NDFU7LT7BTHDNNGYV54MVJJVSD","merge_version":"pith-open-graph-merge-v1","event_count":2,"valid_event_count":2,"invalid_event_count":0,"equivocation_count":0,"current":{"canonical_record":{"metadata":{"abstract_canon_sha256":"9e4cdcaee995b076a8ff0dec95a8c56b9d0790d1ac75b6b49333f09945cb0f66","cross_cats_sorted":[],"license":"http://creativecommons.org/licenses/by-nc-sa/4.0/","primary_cat":"cs.CV","submitted_at":"2026-05-13T11:53:19Z","title_canon_sha256":"f5d0e16b63a6af67a943f55eefeeb4a9ba81386f2424d207404c05bfb85ba27d"},"schema_version":"1.0","source":{"id":"2605.13396","kind":"arxiv","version":1}},"source_aliases":[{"alias_kind":"arxiv","alias_value":"2605.13396","created_at":"2026-05-18T02:44:47Z"},{"alias_kind":"arxiv_version","alias_value":"2605.13396v1","created_at":"2026-05-18T02:44:47Z"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2605.13396","created_at":"2026-05-18T02:44:47Z"},{"alias_kind":"pith_short_12","alias_value":"NDFU7LT7BTHD","created_at":"2026-05-18T12:33:37Z"},{"alias_kind":"pith_short_16","alias_value":"NDFU7LT7BTHDNNGY","created_at":"2026-05-18T12:33:37Z"},{"alias_kind":"pith_short_8","alias_value":"NDFU7LT7","created_at":"2026-05-18T12:33:37Z"}],"graph_snapshots":[{"event_id":"sha256:bd7858d65bedb46387d0af5db8a6603c84596ab1d0d9a580ab867c7ffe6d4773","target":"graph","created_at":"2026-05-18T02:44:47Z","signer":{"key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signer_id":"pith.science","signer_type":"pith_registry"},"payload":{"graph_snapshot":{"author_claims":{"count":0,"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57","strong_count":0},"builder_version":"pith-number-builder-2026-05-17-v1","claims":{"count":4,"items":[{"attestation":"unclaimed","claim_id":"C1","kind":"strongest_claim","source":"verdict.strongest_claim","status":"machine_extracted","text":"PreFIQs quantifies image utility as the Euclidean distance between L2-normalized embeddings extracted from a pre-trained FR model and its pruned counterpart... achieves competitive or superior performance compared to state-of-the-art FIQA methods, including establishing new state-of-the-art results on several benchmarks, without any training or supervision."},{"attestation":"unclaimed","claim_id":"C2","kind":"weakest_assumption","source":"verdict.weakest_assumption","status":"machine_extracted","text":"We hypothesize that low-utility face images rely disproportionately on fragile network parameters, resulting in larger geometric displacement of their embeddings under model sparsification."},{"attestation":"unclaimed","claim_id":"C3","kind":"one_line_summary","source":"verdict.one_line_summary","status":"machine_extracted","text":"Face image quality is quantified as the Euclidean distance between embeddings from a pre-trained face recognition model and its pruned version, achieving competitive or superior results without training or supervision."},{"attestation":"unclaimed","claim_id":"C4","kind":"headline","source":"verdict.pith_extraction.headline","status":"machine_extracted","text":"Face image quality equals the embedding shift that occurs when a face recognition model is pruned."}],"snapshot_sha256":"a220ca0bccfa0dd97d5b59f9bf5d9572e5f76f30a8bae47ca9e4d29ba7ea15e2"},"formal_canon":{"evidence_count":0,"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"paper":{"abstract_excerpt":"Face Image Quality Assessment (FIQA) evaluates the utility of a face image for automated face recognition (FR) systems. In this work, we propose PreFIQs, an unsupervised and training-free FIQA framework grounded in the Pruning Identified Exemplar (PIE) hypothesis. We hypothesize that low-utility face images rely disproportionately on fragile network parameters, resulting in larger geometric displacement of their embeddings under model sparsification. Accordingly, PreFIQs quantifies image utility as the Euclidean distance between L2-normalized embeddings extracted from a pre-trained FR model an","authors_text":"Andrea Atzori, Fadi Boutros, Guray Ozgur, Jan Niklas Kolf, Naser Damer, Vitomir \\v{S}truc, \\v{Z}iga Babnik","cross_cats":[],"headline":"Face image quality equals the embedding shift that occurs when a face recognition model is pruned.","license":"http://creativecommons.org/licenses/by-nc-sa/4.0/","primary_cat":"cs.CV","submitted_at":"2026-05-13T11:53:19Z","title":"PreFIQs: Face Image Quality Is What Survives Pruning"},"references":{"count":56,"internal_anchors":2,"resolved_work":56,"sample":[{"cited_arxiv_id":"","doi":"","is_internal_anchor":false,"ref_index":1,"title":"Deep network pruning: A comparative study on cnns in face recognition","work_id":"fc28e841-f615-4c77-8542-37f07d28c01b","year":2025},{"cited_arxiv_id":"","doi":"","is_internal_anchor":false,"ref_index":2,"title":"Vit- fiqa: Assessing face image quality using vision trans- formers","work_id":"0869dcd4-8792-43a0-8981-8f6a211d5956","year":2025},{"cited_arxiv_id":"","doi":"","is_internal_anchor":false,"ref_index":3,"title":"Faceqan: Face image quality assessment through adversarial noise exploration","work_id":"0417d38a-ab6b-4015-ac1c-4d322b5b1bc0","year":2022},{"cited_arxiv_id":"","doi":"","is_internal_anchor":false,"ref_index":4,"title":"Diffiqa: Face image quality assessment using denoising dif- fusion probabilistic models","work_id":"7c65940e-2dfb-4c9d-bc32-96a2ab943867","year":2023},{"cited_arxiv_id":"","doi":"","is_internal_anchor":false,"ref_index":5,"title":"eDifFIQA: Towards Efficient Face Image Quality Assessment based on Denoising Diffusion Probabilistic Models","work_id":"e5c985eb-0296-4e03-9097-f1b2a0633046","year":2024}],"snapshot_sha256":"bfa0e7d8ea8a01502e9a30f3290f9909ddf7e56589d667e2f5d2249cd5199bf2"},"source":{"id":"2605.13396","kind":"arxiv","version":1},"verdict":{"created_at":"2026-05-14T20:46:31.142595Z","id":"92c26ae3-2e6d-4d5d-9cf3-fee4be08eb71","model_set":{"reader":"grok-4.3"},"one_line_summary":"Face image quality is quantified as the Euclidean distance between embeddings from a pre-trained face recognition model and its pruned version, achieving competitive or superior results without training or supervision.","pipeline_version":"pith-pipeline@v0.9.0","pith_extraction_headline":"Face image quality equals the embedding shift that occurs when a face recognition model is pruned.","strongest_claim":"PreFIQs quantifies image utility as the Euclidean distance between L2-normalized embeddings extracted from a pre-trained FR model and its pruned counterpart... achieves competitive or superior performance compared to state-of-the-art FIQA methods, including establishing new state-of-the-art results on several benchmarks, without any training or supervision.","weakest_assumption":"We hypothesize that low-utility face images rely disproportionately on fragile network parameters, resulting in larger geometric displacement of their embeddings under model sparsification."}},"verdict_id":"92c26ae3-2e6d-4d5d-9cf3-fee4be08eb71"}}],"author_attestations":[],"timestamp_anchors":[],"storage_attestations":[],"citation_signatures":[],"replication_records":[],"corrections":[],"mirror_hints":[],"record_created":{"event_id":"sha256:c77a9dea1207fd7e4cad7d596d62159ebe1d59d7fa43b5c38650b90e3c9a9c03","target":"record","created_at":"2026-05-18T02:44:47Z","signer":{"key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signer_id":"pith.science","signer_type":"pith_registry"},"payload":{"attestation_state":"computed","canonical_record":{"metadata":{"abstract_canon_sha256":"9e4cdcaee995b076a8ff0dec95a8c56b9d0790d1ac75b6b49333f09945cb0f66","cross_cats_sorted":[],"license":"http://creativecommons.org/licenses/by-nc-sa/4.0/","primary_cat":"cs.CV","submitted_at":"2026-05-13T11:53:19Z","title_canon_sha256":"f5d0e16b63a6af67a943f55eefeeb4a9ba81386f2424d207404c05bfb85ba27d"},"schema_version":"1.0","source":{"id":"2605.13396","kind":"arxiv","version":1}},"canonical_sha256":"68cb4fae7f0cce36b4d8af78caa53590e46729fd458e07a8b3a715fb575cd129","receipt":{"algorithm":"ed25519","builder_version":"pith-number-builder-2026-05-17-v1","canonical_sha256":"68cb4fae7f0cce36b4d8af78caa53590e46729fd458e07a8b3a715fb575cd129","first_computed_at":"2026-05-18T02:44:47.651886Z","key_id":"pith-v1-2026-05","kind":"pith_receipt","last_reissued_at":"2026-05-18T02:44:47.651886Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","receipt_version":"0.3","signature_b64":"kd7vKcYtbrCc86VeUmMHpGqU0tC/I7KgfuZv/UCOrqJSsS29a6b5C0fHz1APWsI+kt/tDxm9zlh3osiNSChiDQ==","signature_status":"signed_v1","signed_at":"2026-05-18T02:44:47.652351Z","signed_message":"canonical_sha256_bytes"},"source_id":"2605.13396","source_kind":"arxiv","source_version":1}}},"equivocations":[],"invalid_events":[],"applied_event_ids":["sha256:c77a9dea1207fd7e4cad7d596d62159ebe1d59d7fa43b5c38650b90e3c9a9c03","sha256:bd7858d65bedb46387d0af5db8a6603c84596ab1d0d9a580ab867c7ffe6d4773"],"state_sha256":"44c5188354001bcdcee3094df6cd40e8d233e627c00928240dec4a221816f486"},"bundle_signature":{"signature_status":"signed_v1","algorithm":"ed25519","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signature_b64":"n+lSxxny2SIefESUuil4jpvif9D2KMFBP/KzOEtp8TqItbcgwnV0IpOAOe96agOeuWQjBQu99Pa2+bWSFKTWAg==","signed_message":"bundle_sha256_bytes","signed_at":"2026-06-06T19:20:11.885333Z","bundle_sha256":"e189d7b9cbe60933edcfd028179db201b6d84ab7dd1b5b6cff52013277d84968"}}