{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2022:N6BQEGEBRY76ECFZMFMW6ACT4L","short_pith_number":"pith:N6BQEGEB","schema_version":"1.0","canonical_sha256":"6f830218818e3fe208b961596f0053e2f87fa7e98a10268ee4f301534de1d88f","source":{"kind":"arxiv","id":"2210.04150","version":3},"attestation_state":"computed","paper":{"title":"Open-Vocabulary Semantic Segmentation with Mask-adapted CLIP","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":["cs.LG"],"primary_cat":"cs.CV","authors_text":"Bichen Wu, Diana Marculescu, Feng Liang, Hang Zhang, Kunpeng Li, Peizhao Zhang, Peter Vajda, Xiaoliang Dai, Yinan Zhao","submitted_at":"2022-10-09T02:57:32Z","abstract_excerpt":"Open-vocabulary semantic segmentation aims to segment an image into semantic regions according to text descriptions, which may not have been seen during training. Recent two-stage methods first generate class-agnostic mask proposals and then leverage pre-trained vision-language models, e.g., CLIP, to classify masked regions. We identify the performance bottleneck of this paradigm to be the pre-trained CLIP model, since it does not perform well on masked images. To address this, we propose to finetune CLIP on a collection of masked image regions and their corresponding text descriptions. We col"},"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":"2210.04150","kind":"arxiv","version":3},"metadata":{"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.CV","submitted_at":"2022-10-09T02:57:32Z","cross_cats_sorted":["cs.LG"],"title_canon_sha256":"ab5e61143bfc259f5649667d38bdc36db4845ebc76d4f28f7a57fd47ae138a0d","abstract_canon_sha256":"61aa96df243ab1dfa327f09a525430978b6ce1a7c54bbce94ee3464805499422"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T05:56:54.845042Z","signature_b64":"U1WZduw0Aba2f1/CI8gN2o9HnD8/WnsCrkLVsEJYRQPXcTPAqQ1qgGkdV8TNy7Fd0qhNsypXmdIyk0ORs15/DA==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"6f830218818e3fe208b961596f0053e2f87fa7e98a10268ee4f301534de1d88f","last_reissued_at":"2026-07-05T05:56:54.844558Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T05:56:54.844558Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"Open-Vocabulary Semantic Segmentation with Mask-adapted CLIP","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":["cs.LG"],"primary_cat":"cs.CV","authors_text":"Bichen Wu, Diana Marculescu, Feng Liang, Hang Zhang, Kunpeng Li, Peizhao Zhang, Peter Vajda, Xiaoliang Dai, Yinan Zhao","submitted_at":"2022-10-09T02:57:32Z","abstract_excerpt":"Open-vocabulary semantic segmentation aims to segment an image into semantic regions according to text descriptions, which may not have been seen during training. Recent two-stage methods first generate class-agnostic mask proposals and then leverage pre-trained vision-language models, e.g., CLIP, to classify masked regions. We identify the performance bottleneck of this paradigm to be the pre-trained CLIP model, since it does not perform well on masked images. To address this, we propose to finetune CLIP on a collection of masked image regions and their corresponding text descriptions. We col"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2210.04150","kind":"arxiv","version":3},"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/2210.04150/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":"2210.04150","created_at":"2026-07-05T05:56:54.844622+00:00"},{"alias_kind":"arxiv_version","alias_value":"2210.04150v3","created_at":"2026-07-05T05:56:54.844622+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2210.04150","created_at":"2026-07-05T05:56:54.844622+00:00"},{"alias_kind":"pith_short_12","alias_value":"N6BQEGEBRY76","created_at":"2026-07-05T05:56:54.844622+00:00"},{"alias_kind":"pith_short_16","alias_value":"N6BQEGEBRY76ECFZ","created_at":"2026-07-05T05:56:54.844622+00:00"},{"alias_kind":"pith_short_8","alias_value":"N6BQEGEB","created_at":"2026-07-05T05:56:54.844622+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":1,"internal_anchor_count":1,"sample":[{"citing_arxiv_id":"2509.01275","citing_title":"Novel Category Discovery with X-Agent Attention for Open-Vocabulary Semantic Segmentation","ref_index":26,"is_internal_anchor":true}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/N6BQEGEBRY76ECFZMFMW6ACT4L","json":"https://pith.science/pith/N6BQEGEBRY76ECFZMFMW6ACT4L.json","graph_json":"https://pith.science/api/pith-number/N6BQEGEBRY76ECFZMFMW6ACT4L/graph.json","events_json":"https://pith.science/api/pith-number/N6BQEGEBRY76ECFZMFMW6ACT4L/events.json","paper":"https://pith.science/paper/N6BQEGEB"},"agent_actions":{"view_html":"https://pith.science/pith/N6BQEGEBRY76ECFZMFMW6ACT4L","download_json":"https://pith.science/pith/N6BQEGEBRY76ECFZMFMW6ACT4L.json","view_paper":"https://pith.science/paper/N6BQEGEB","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2210.04150&json=true","fetch_graph":"https://pith.science/api/pith-number/N6BQEGEBRY76ECFZMFMW6ACT4L/graph.json","fetch_events":"https://pith.science/api/pith-number/N6BQEGEBRY76ECFZMFMW6ACT4L/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/N6BQEGEBRY76ECFZMFMW6ACT4L/action/timestamp_anchor","attest_storage":"https://pith.science/pith/N6BQEGEBRY76ECFZMFMW6ACT4L/action/storage_attestation","attest_author":"https://pith.science/pith/N6BQEGEBRY76ECFZMFMW6ACT4L/action/author_attestation","sign_citation":"https://pith.science/pith/N6BQEGEBRY76ECFZMFMW6ACT4L/action/citation_signature","submit_replication":"https://pith.science/pith/N6BQEGEBRY76ECFZMFMW6ACT4L/action/replication_record"}},"created_at":"2026-07-05T05:56:54.844622+00:00","updated_at":"2026-07-05T05:56:54.844622+00:00"}