{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2026:26GH5WORHYUEK372LHSES7T333","short_pith_number":"pith:26GH5WOR","schema_version":"1.0","canonical_sha256":"d78c7ed9d13e28456ffa59e4497e7bdeeeaf92c0a6f92aebf01b29f637e2897c","source":{"kind":"arxiv","id":"2605.15542","version":1},"attestation_state":"computed","paper":{"title":"DRS-GUI: Dynamic Region Search for Training-Free GUI Grounding","license":"http://creativecommons.org/licenses/by/4.0/","headline":"A training-free dynamic region search method improves GUI grounding performance by 14 percent in existing multimodal models.","cross_cats":[],"primary_cat":"cs.AI","authors_text":"Huawen Shen, Liu Yu, Shiyu Liu, Yichao Liu, Yu Zhou, Zeyu Chen","submitted_at":"2026-05-15T02:27:41Z","abstract_excerpt":"GUI agents powered by Multimodal Large Language Models (MLLMs) have demonstrated impressive capability in understanding and executing user instructions. However, accurately grounding instruction-relevant elements from high-resolution screenshots cluttered with irrelevant UI components remains challenging for existing approaches. Inspired by how humans dynamically adjust their perceptual scope to locate task-related regions on complex screens, we propose DRS-GUI, a training-free dynamic region search framework for GUI grounding that can be seamlessly integrated into existing MLLMs. DRS-GUI intr"},"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":true,"formal_links_present":true},"canonical_record":{"source":{"id":"2605.15542","kind":"arxiv","version":1},"metadata":{"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.AI","submitted_at":"2026-05-15T02:27:41Z","cross_cats_sorted":[],"title_canon_sha256":"72b5086792ce21d0a8928bc06558813c669fe1b879e949a1205315e775d4f736","abstract_canon_sha256":"1505f23c909af627f22c2459b06139f2a518910bb0d28529294ed56f6b4fb403"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-05-20T00:01:04.401361Z","signature_b64":"Ly4dmNMnQ3gFONMKI9ZR5eik/CwfII3B+hhcBBSWyg51K1zK3YitPA3Gqv2+/0mUYWMQs91YAwQ/E09Zfjw9Dw==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"d78c7ed9d13e28456ffa59e4497e7bdeeeaf92c0a6f92aebf01b29f637e2897c","last_reissued_at":"2026-05-20T00:01:04.400686Z","signature_status":"signed_v1","first_computed_at":"2026-05-20T00:01:04.400686Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"DRS-GUI: Dynamic Region Search for Training-Free GUI Grounding","license":"http://creativecommons.org/licenses/by/4.0/","headline":"A training-free dynamic region search method improves GUI grounding performance by 14 percent in existing multimodal models.","cross_cats":[],"primary_cat":"cs.AI","authors_text":"Huawen Shen, Liu Yu, Shiyu Liu, Yichao Liu, Yu Zhou, Zeyu Chen","submitted_at":"2026-05-15T02:27:41Z","abstract_excerpt":"GUI agents powered by Multimodal Large Language Models (MLLMs) have demonstrated impressive capability in understanding and executing user instructions. However, accurately grounding instruction-relevant elements from high-resolution screenshots cluttered with irrelevant UI components remains challenging for existing approaches. Inspired by how humans dynamically adjust their perceptual scope to locate task-related regions on complex screens, we propose DRS-GUI, a training-free dynamic region search framework for GUI grounding that can be seamlessly integrated into existing MLLMs. DRS-GUI intr"},"claims":{"count":4,"items":[{"kind":"strongest_claim","text":"Experiments demonstrate that DRS-GUI yields a 14% improvement on ScreenSpot-Pro for general and GUI-specific MLLMs (Qwen2.5-VL-7B and UGround-V1-7B), significantly enhancing grounding performance and generalization.","source":"verdict.strongest_claim","status":"machine_extracted","claim_id":"C1","attestation":"unclaimed"},{"kind":"weakest_assumption","text":"The three perceptual actions (Focus, Shift, and Scatter) performed by the lightweight UI Perceptor, when scheduled by the MCTS-based Action Planner, will reliably generate and select instruction-relevant region proposals from cluttered high-resolution screenshots without requiring any model training or fine-tuning.","source":"verdict.weakest_assumption","status":"machine_extracted","claim_id":"C2","attestation":"unclaimed"},{"kind":"one_line_summary","text":"DRS-GUI introduces a dynamic region search method with Focus/Shift/Scatter actions and MCTS-based planning that improves GUI grounding accuracy by 14% on ScreenSpot-Pro for both general and GUI-specific MLLMs without any training.","source":"verdict.one_line_summary","status":"machine_extracted","claim_id":"C3","attestation":"unclaimed"},{"kind":"headline","text":"A training-free dynamic region search method improves GUI grounding performance by 14 percent in existing multimodal models.","source":"verdict.pith_extraction.headline","status":"machine_extracted","claim_id":"C4","attestation":"unclaimed"}],"snapshot_sha256":"c61e82e9070dc21b7ac5c50d7dd27a591462b53c6dbe2a0079a320ed72ed6c66"},"source":{"id":"2605.15542","kind":"arxiv","version":1},"verdict":{"id":"7dc9aa6b-3082-4e00-9a7e-ffab5cf2943a","model_set":{"reader":"grok-4.3"},"created_at":"2026-05-19T14:20:03.742235Z","strongest_claim":"Experiments demonstrate that DRS-GUI yields a 14% improvement on ScreenSpot-Pro for general and GUI-specific MLLMs (Qwen2.5-VL-7B and UGround-V1-7B), significantly enhancing grounding performance and generalization.","one_line_summary":"DRS-GUI introduces a dynamic region search method with Focus/Shift/Scatter actions and MCTS-based planning that improves GUI grounding accuracy by 14% on ScreenSpot-Pro for both general and GUI-specific MLLMs without any training.","pipeline_version":"pith-pipeline@v0.9.0","weakest_assumption":"The three perceptual actions (Focus, Shift, and Scatter) performed by the lightweight UI Perceptor, when scheduled by the MCTS-based Action Planner, will reliably generate and select instruction-relevant region proposals from cluttered high-resolution screenshots without requiring any model training or fine-tuning.","pith_extraction_headline":"A training-free dynamic region search method improves GUI grounding performance by 14 percent in existing multimodal models."},"integrity":{"clean":true,"summary":{"advisory":0,"critical":0,"by_detector":{},"informational":0},"endpoint":"/pith/2605.15542/integrity.json","findings":[],"available":true,"detectors_run":[{"name":"doi_title_agreement","ran_at":"2026-05-19T14:31:17.431374Z","status":"completed","version":"1.0.0","findings_count":0},{"name":"doi_compliance","ran_at":"2026-05-19T14:27:26.780987Z","status":"completed","version":"1.0.0","findings_count":0},{"name":"cited_work_retraction","ran_at":"2026-05-19T14:22:00.981090Z","status":"completed","version":"1.0.0","findings_count":0},{"name":"claim_evidence","ran_at":"2026-05-19T14:21:54.025480Z","status":"completed","version":"1.0.0","findings_count":0},{"name":"shingle_duplication","ran_at":"2026-05-19T13:49:41.825869Z","status":"skipped","version":"0.1.0","findings_count":0},{"name":"citation_quote_validity","ran_at":"2026-05-19T13:49:41.363442Z","status":"skipped","version":"0.1.0","findings_count":0},{"name":"ai_meta_artifact","ran_at":"2026-05-19T13:33:22.611102Z","status":"skipped","version":"1.0.0","findings_count":0}],"snapshot_sha256":"190d8ee564ffea5cfd7a5b068ab5e849c38054971c89f8f1f943d7f3302ab476"},"references":{"count":54,"sample":[{"doi":"","year":null,"title":"GPT-4 Technical Report","work_id":"b928e041-6991-4c08-8c81-0359e4097c7b","ref_index":1,"cited_arxiv_id":"2303.08774","is_internal_anchor":true},{"doi":"","year":2023,"title":"Qwen Technical Report","work_id":"bb1fd52f-6b2f-437c-9516-37bdf6eb9be8","ref_index":2,"cited_arxiv_id":"2309.16609","is_internal_anchor":true},{"doi":"","year":2023,"title":"Qwen-VL: A Versatile Vision-Language Model for Understanding, Localization, Text Reading, and Beyond","work_id":"cbc2bb21-b6bb-46c0-80bf-107e195ffe10","ref_index":3,"cited_arxiv_id":"2308.12966","is_internal_anchor":true},{"doi":"","year":2025,"title":"Qwen2.5-VL Technical Report","work_id":"69dffacb-bfe8-442d-be86-48624c60426f","ref_index":4,"cited_arxiv_id":"2502.13923","is_internal_anchor":true},{"doi":"","year":2023,"title":"Introducing our multimodal models, 2023","work_id":"5e64299f-ba20-4018-9460-81ed6cca321b","ref_index":5,"cited_arxiv_id":"","is_internal_anchor":false}],"resolved_work":54,"snapshot_sha256":"85d9a02f82f3edb09c8ed74c6989a51f357399345983fbeca897831fd43cd0cc","internal_anchors":17},"formal_canon":{"evidence_count":2,"snapshot_sha256":"1d3f1858a8a791bb625e3df1449a4e969a8d75fd2c50c6dacc74869c4075ee2a"},"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":"2605.15542","created_at":"2026-05-20T00:01:04.400812+00:00"},{"alias_kind":"arxiv_version","alias_value":"2605.15542v1","created_at":"2026-05-20T00:01:04.400812+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2605.15542","created_at":"2026-05-20T00:01:04.400812+00:00"},{"alias_kind":"pith_short_12","alias_value":"26GH5WORHYUE","created_at":"2026-05-20T00:01:04.400812+00:00"},{"alias_kind":"pith_short_16","alias_value":"26GH5WORHYUEK372","created_at":"2026-05-20T00:01:04.400812+00:00"},{"alias_kind":"pith_short_8","alias_value":"26GH5WOR","created_at":"2026-05-20T00:01:04.400812+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":0,"internal_anchor_count":0,"sample":[]},"formal_canon":{"evidence_count":2,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/26GH5WORHYUEK372LHSES7T333","json":"https://pith.science/pith/26GH5WORHYUEK372LHSES7T333.json","graph_json":"https://pith.science/api/pith-number/26GH5WORHYUEK372LHSES7T333/graph.json","events_json":"https://pith.science/api/pith-number/26GH5WORHYUEK372LHSES7T333/events.json","paper":"https://pith.science/paper/26GH5WOR"},"agent_actions":{"view_html":"https://pith.science/pith/26GH5WORHYUEK372LHSES7T333","download_json":"https://pith.science/pith/26GH5WORHYUEK372LHSES7T333.json","view_paper":"https://pith.science/paper/26GH5WOR","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2605.15542&json=true","fetch_graph":"https://pith.science/api/pith-number/26GH5WORHYUEK372LHSES7T333/graph.json","fetch_events":"https://pith.science/api/pith-number/26GH5WORHYUEK372LHSES7T333/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/26GH5WORHYUEK372LHSES7T333/action/timestamp_anchor","attest_storage":"https://pith.science/pith/26GH5WORHYUEK372LHSES7T333/action/storage_attestation","attest_author":"https://pith.science/pith/26GH5WORHYUEK372LHSES7T333/action/author_attestation","sign_citation":"https://pith.science/pith/26GH5WORHYUEK372LHSES7T333/action/citation_signature","submit_replication":"https://pith.science/pith/26GH5WORHYUEK372LHSES7T333/action/replication_record"}},"created_at":"2026-05-20T00:01:04.400812+00:00","updated_at":"2026-05-20T00:01:04.400812+00:00"}