{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2026:IQGO64UQFB4GBW2NCOSR4VVWLE","short_pith_number":"pith:IQGO64UQ","schema_version":"1.0","canonical_sha256":"440cef7290287860db4d13a51e56b65938a69ca578a33177259a65464e164923","source":{"kind":"arxiv","id":"2605.22132","version":1},"attestation_state":"computed","paper":{"title":"Accelerating Vision Foundation Models with Drop-in Depthwise Convolution","license":"http://creativecommons.org/licenses/by-nc-sa/4.0/","headline":"","cross_cats":[],"primary_cat":"cs.CV","authors_text":"Carmelo Scribano, Danda Pani Paudel, Giorgia Franchini, Luc Van Gool, Marko Bertogna, Mohammad Mahdi, Nedyalko Prisadnikov, Yuqian Fu","submitted_at":"2026-05-21T08:07:23Z","abstract_excerpt":"Pretrained vision foundation models deliver strong performance across tasks with limited fine-tuning. However, their Vision Transformer (ViT) backbones impose high inference costs, limiting deployment on resource-constrained devices. In this work, we accelerate large-scale pretrained ViTs while preserving their feature extraction capabilities by exploiting the intrinsic convolution-like behavior of some attention heads. Specifically, we introduce an efficient depthwise convolution-based layer that serves as a drop-in replacement for these heads. Additionally, we propose simple strategies to id"},"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":"2605.22132","kind":"arxiv","version":1},"metadata":{"license":"http://creativecommons.org/licenses/by-nc-sa/4.0/","primary_cat":"cs.CV","submitted_at":"2026-05-21T08:07:23Z","cross_cats_sorted":[],"title_canon_sha256":"ebf49f7493b777f1a782a8bf00bf3474478489e6110e8ee1191b0aa7568952fa","abstract_canon_sha256":"6459432e413d6493c64d68d730c6fa5dde0346fc6fba4cbe7001fc9bbaac030c"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-05-22T01:04:27.466674Z","signature_b64":"VAVz/5oqGT6bJ0gKHNLDXJO+tF+HzEapMXn2uWjs9xgyyS1cCE4/VYeoAz0FOwxuI/hsmx4odI5eit7/jJKBCA==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"440cef7290287860db4d13a51e56b65938a69ca578a33177259a65464e164923","last_reissued_at":"2026-05-22T01:04:27.465949Z","signature_status":"signed_v1","first_computed_at":"2026-05-22T01:04:27.465949Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"Accelerating Vision Foundation Models with Drop-in Depthwise Convolution","license":"http://creativecommons.org/licenses/by-nc-sa/4.0/","headline":"","cross_cats":[],"primary_cat":"cs.CV","authors_text":"Carmelo Scribano, Danda Pani Paudel, Giorgia Franchini, Luc Van Gool, Marko Bertogna, Mohammad Mahdi, Nedyalko Prisadnikov, Yuqian Fu","submitted_at":"2026-05-21T08:07:23Z","abstract_excerpt":"Pretrained vision foundation models deliver strong performance across tasks with limited fine-tuning. However, their Vision Transformer (ViT) backbones impose high inference costs, limiting deployment on resource-constrained devices. In this work, we accelerate large-scale pretrained ViTs while preserving their feature extraction capabilities by exploiting the intrinsic convolution-like behavior of some attention heads. Specifically, we introduce an efficient depthwise convolution-based layer that serves as a drop-in replacement for these heads. Additionally, we propose simple strategies to id"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2605.22132","kind":"arxiv","version":1},"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/2605.22132/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":"2605.22132","created_at":"2026-05-22T01:04:27.466083+00:00"},{"alias_kind":"arxiv_version","alias_value":"2605.22132v1","created_at":"2026-05-22T01:04:27.466083+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2605.22132","created_at":"2026-05-22T01:04:27.466083+00:00"},{"alias_kind":"pith_short_12","alias_value":"IQGO64UQFB4G","created_at":"2026-05-22T01:04:27.466083+00:00"},{"alias_kind":"pith_short_16","alias_value":"IQGO64UQFB4GBW2N","created_at":"2026-05-22T01:04:27.466083+00:00"},{"alias_kind":"pith_short_8","alias_value":"IQGO64UQ","created_at":"2026-05-22T01:04:27.466083+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":0,"internal_anchor_count":0,"sample":[]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/IQGO64UQFB4GBW2NCOSR4VVWLE","json":"https://pith.science/pith/IQGO64UQFB4GBW2NCOSR4VVWLE.json","graph_json":"https://pith.science/api/pith-number/IQGO64UQFB4GBW2NCOSR4VVWLE/graph.json","events_json":"https://pith.science/api/pith-number/IQGO64UQFB4GBW2NCOSR4VVWLE/events.json","paper":"https://pith.science/paper/IQGO64UQ"},"agent_actions":{"view_html":"https://pith.science/pith/IQGO64UQFB4GBW2NCOSR4VVWLE","download_json":"https://pith.science/pith/IQGO64UQFB4GBW2NCOSR4VVWLE.json","view_paper":"https://pith.science/paper/IQGO64UQ","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2605.22132&json=true","fetch_graph":"https://pith.science/api/pith-number/IQGO64UQFB4GBW2NCOSR4VVWLE/graph.json","fetch_events":"https://pith.science/api/pith-number/IQGO64UQFB4GBW2NCOSR4VVWLE/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/IQGO64UQFB4GBW2NCOSR4VVWLE/action/timestamp_anchor","attest_storage":"https://pith.science/pith/IQGO64UQFB4GBW2NCOSR4VVWLE/action/storage_attestation","attest_author":"https://pith.science/pith/IQGO64UQFB4GBW2NCOSR4VVWLE/action/author_attestation","sign_citation":"https://pith.science/pith/IQGO64UQFB4GBW2NCOSR4VVWLE/action/citation_signature","submit_replication":"https://pith.science/pith/IQGO64UQFB4GBW2NCOSR4VVWLE/action/replication_record"}},"created_at":"2026-05-22T01:04:27.466083+00:00","updated_at":"2026-05-22T01:04:27.466083+00:00"}