{"paper":{"title":"Geometry Reinforced Efficient Attention Tuning Equipped with Normals for Robust Stereo Matching","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"Surface normals provide domain-invariant geometric cues that improve zero-shot generalization in stereo matching from synthetic to real data.","cross_cats":[],"primary_cat":"cs.CV","authors_text":"Cheng Huang, Jiahao Li, Jianping Wang, Xinhong Chen, Yung-Hui Li, Zhengmin Jiang","submitted_at":"2026-04-10T09:21:51Z","abstract_excerpt":"Despite remarkable advances in image-driven stereo matching over the past decade, Synthetic-to-Realistic ZeroShot (Syn-to-Real) generalization remains an open challenge. This suboptimal generalization performance mainly stems from cross-domain shifts and ill-posed ambiguities inherent in image textures, particularly in occluded, textureless, repetitive, and non-Lambertian (specular/transparent) regions. To improve Synto-Real generalization, we propose GREATEN, a framework that incorporates surface normals as domain-invariant, object-intrinsic, and discriminative geometric cues to compensate fo"},"claims":{"count":4,"items":[{"kind":"strongest_claim","text":"Trained exclusively on synthetic data such as SceneFlow, GREATEN-IGEV achieves outstanding Syn-to-Real performance. Specifically, it reduces errors by 30% on ETH3D, 8.5% on the non-Lambertian Booster, and 14.1% on KITTI-2015, compared to FoundationStereo, Monster-Stereo, and DEFOM-Stereo, respectively.","source":"verdict.strongest_claim","status":"machine_extracted","claim_id":"C1","attestation":"unclaimed"},{"kind":"weakest_assumption","text":"That surface normals can be obtained or estimated reliably enough in real scenes to serve as domain-invariant cues, and that the gated fusion module will consistently suppress misleading image textures without introducing new artifacts in occluded or non-Lambertian regions.","source":"verdict.weakest_assumption","status":"machine_extracted","claim_id":"C2","attestation":"unclaimed"},{"kind":"one_line_summary","text":"GREATEN fuses surface normals with image features via gated contextual-geometric fusion and efficient sparse attentions to cut stereo matching errors by up to 30% on real datasets when trained solely on synthetic data.","source":"verdict.one_line_summary","status":"machine_extracted","claim_id":"C3","attestation":"unclaimed"},{"kind":"headline","text":"Surface normals provide domain-invariant geometric cues that improve zero-shot generalization in stereo matching from synthetic to real data.","source":"verdict.pith_extraction.headline","status":"machine_extracted","claim_id":"C4","attestation":"unclaimed"}],"snapshot_sha256":"fc2beb080f28c46d056b3d9c91212495e847e24950b40b137cf2c73bd224ac3b"},"source":{"id":"2604.09142","kind":"arxiv","version":2},"verdict":{"id":"b7018f9d-3b16-41dd-af5c-9b5fe4b3851b","model_set":{"reader":"grok-4.3"},"created_at":"2026-05-10T17:48:03.616235Z","strongest_claim":"Trained exclusively on synthetic data such as SceneFlow, GREATEN-IGEV achieves outstanding Syn-to-Real performance. Specifically, it reduces errors by 30% on ETH3D, 8.5% on the non-Lambertian Booster, and 14.1% on KITTI-2015, compared to FoundationStereo, Monster-Stereo, and DEFOM-Stereo, respectively.","one_line_summary":"GREATEN fuses surface normals with image features via gated contextual-geometric fusion and efficient sparse attentions to cut stereo matching errors by up to 30% on real datasets when trained solely on synthetic data.","pipeline_version":"pith-pipeline@v0.9.0","weakest_assumption":"That surface normals can be obtained or estimated reliably enough in real scenes to serve as domain-invariant cues, and that the gated fusion module will consistently suppress misleading image textures without introducing new artifacts in occluded or non-Lambertian regions.","pith_extraction_headline":"Surface normals provide domain-invariant geometric cues that improve zero-shot generalization in stereo matching from synthetic to real data."},"integrity":{"clean":true,"summary":{"advisory":0,"critical":0,"by_detector":{},"informational":0},"endpoint":"/pith/2604.09142/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"}