{"paper":{"title":"Faster Segment Anything: Towards Lightweight SAM for Mobile Applications","license":"http://creativecommons.org/licenses/by-nc-sa/4.0/","headline":"Distilling SAM's heavy encoder into a lightweight one creates MobileSAM, over 60 times smaller with matching zero-shot segmentation performance.","cross_cats":[],"primary_cat":"cs.CV","authors_text":"Chaoning Zhang, Choong Seon Hong, Dongshen Han, Jung Uk Kim, Seungkyu Lee, Sung-Ho Bae, Yu Qiao","submitted_at":"2023-06-25T16:37:25Z","abstract_excerpt":"Segment Anything Model (SAM) has attracted significant attention due to its impressive zero-shot transfer performance and high versatility for numerous vision applications (like image editing with fine-grained control). Many of such applications need to be run on resource-constraint edge devices, like mobile phones. In this work, we aim to make SAM mobile-friendly by replacing the heavyweight image encoder with a lightweight one. A naive way to train such a new SAM as in the original SAM paper leads to unsatisfactory performance, especially when limited training sources are available. We find "},"claims":{"count":4,"items":[{"kind":"strongest_claim","text":"the resulting lightweight SAM is termed MobileSAM which is more than 60 times smaller yet performs on par with the original SAM","source":"verdict.strongest_claim","status":"machine_extracted","claim_id":"C1","attestation":"unclaimed"},{"kind":"weakest_assumption","text":"that a lightweight encoder distilled only from the frozen original encoder will remain compatible with the original mask decoder across diverse downstream tasks without further joint fine-tuning","source":"verdict.weakest_assumption","status":"machine_extracted","claim_id":"C2","attestation":"unclaimed"},{"kind":"one_line_summary","text":"MobileSAM is a 60x smaller distilled version of SAM that matches original performance and runs 5x faster than concurrent FastSAM while supporting CPU inference.","source":"verdict.one_line_summary","status":"machine_extracted","claim_id":"C3","attestation":"unclaimed"},{"kind":"headline","text":"Distilling SAM's heavy encoder into a lightweight one creates MobileSAM, over 60 times smaller with matching zero-shot segmentation performance.","source":"verdict.pith_extraction.headline","status":"machine_extracted","claim_id":"C4","attestation":"unclaimed"}],"snapshot_sha256":"56fb1bfec08cd2eae39b54bb9581db78395a38832c2c1f23c2ada5d67f8d3d86"},"source":{"id":"2306.14289","kind":"arxiv","version":2},"verdict":{"id":"9db7228b-9ee9-4bbc-84c1-da0bad73d9af","model_set":{"reader":"grok-4.3"},"created_at":"2026-05-17T22:37:57.264741Z","strongest_claim":"the resulting lightweight SAM is termed MobileSAM which is more than 60 times smaller yet performs on par with the original SAM","one_line_summary":"MobileSAM is a 60x smaller distilled version of SAM that matches original performance and runs 5x faster than concurrent FastSAM while supporting CPU inference.","pipeline_version":"pith-pipeline@v0.9.0","weakest_assumption":"that a lightweight encoder distilled only from the frozen original encoder will remain compatible with the original mask decoder across diverse downstream tasks without further joint fine-tuning","pith_extraction_headline":"Distilling SAM's heavy encoder into a lightweight one creates MobileSAM, over 60 times smaller with matching zero-shot segmentation performance."},"references":{"count":18,"sample":[{"doi":"","year":null,"title":"One small step for generative ai, one giant leap for agi: A complete survey on chatgpt in aigc era","work_id":"e0645322-bb71-4701-b9ff-134537e95fe8","ref_index":1,"cited_arxiv_id":"","is_internal_anchor":false},{"doi":"","year":null,"title":"On the Opportunities and Risks of Foundation Models","work_id":"a18039e9-928d-47c9-a836-32656a71bf71","ref_index":2,"cited_arxiv_id":"2108.07258","is_internal_anchor":true},{"doi":"","year":null,"title":"Mp-fedcl: Multi-prototype federated contrastive learning for edge intelligence","work_id":"c0a2ea68-d6a5-40cb-97a0-b9abfac75009","ref_index":3,"cited_arxiv_id":"","is_internal_anchor":false},{"doi":"","year":null,"title":"Fast segment anything","work_id":"feed3d9f-cc9f-42db-90e6-e9ff051cef57","ref_index":4,"cited_arxiv_id":"","is_internal_anchor":false},{"doi":"","year":null,"title":"Segment anything in medical images","work_id":"b64c0fe5-9896-4720-94a0-9f05513ee885","ref_index":5,"cited_arxiv_id":"","is_internal_anchor":false}],"resolved_work":18,"snapshot_sha256":"2a62ff75b5da33cefb09495749782b1555582fb10f7511aa6c0deb8f0e9a575e","internal_anchors":4},"formal_canon":{"evidence_count":2,"snapshot_sha256":"8025b99c4461f43a4c3d613f9306938736a764a247dc06bde4265860b4459920"},"author_claims":{"count":0,"strong_count":0,"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"builder_version":"pith-number-builder-2026-05-17-v1"}