{"paper":{"title":"WD-FQDet: Multispectral Detection Transformer via Wavelet Decomposition and Frequency-aware Query Learning","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"Wavelet decomposition decouples shared low-frequency and specific high-frequency features from infrared and visible images to improve object detection.","cross_cats":[],"primary_cat":"cs.CV","authors_text":"Chunjin Yang, Fanman Meng, Xiwei Zhang, Yiming Xiao","submitted_at":"2026-05-13T14:51:05Z","abstract_excerpt":"Infrared-visible object detection improves detection performance by combining complementary features from multispectral images. Existing backbone-specific and backbone-shared approaches still suffer from the problems of severe bias of modality-shared features and the insufficiency of modality-specific features. To address these issues, we propose a novel detection framework WD-FQDet that explicitly decouples modality-shared and modality-specific information from infrared and visible modalities in the new view of low- and high-frequency domains, allowing fusion strategies tailored to their freq"},"claims":{"count":4,"items":[{"kind":"strongest_claim","text":"we propose a novel detection framework WD-FQDet that explicitly decouples modality-shared and modality-specific information from infrared and visible modalities in the new view of low- and high-frequency domains, allowing fusion strategies tailored to their frequency characteristics... Experimental results on the FLIR, LLVIP, and M3FD datasets demonstrate that WD-FQDet achieves state-of-the-art performance across multiple evaluation metrics.","source":"verdict.strongest_claim","status":"machine_extracted","claim_id":"C1","attestation":"unclaimed"},{"kind":"weakest_assumption","text":"The assumption that wavelet decomposition cleanly separates modality-shared low-frequency features from modality-specific high-frequency features and that the proposed alignment, retention, and query modules will mitigate bias and insufficiency without introducing artifacts or overfitting to the specific datasets.","source":"verdict.weakest_assumption","status":"machine_extracted","claim_id":"C2","attestation":"unclaimed"},{"kind":"one_line_summary","text":"WD-FQDet decouples modality-shared and modality-specific features in infrared-visible images via wavelet-based frequency decomposition and frequency-aware query selection to achieve state-of-the-art detection performance.","source":"verdict.one_line_summary","status":"machine_extracted","claim_id":"C3","attestation":"unclaimed"},{"kind":"headline","text":"Wavelet decomposition decouples shared low-frequency and specific high-frequency features from infrared and visible images to improve object detection.","source":"verdict.pith_extraction.headline","status":"machine_extracted","claim_id":"C4","attestation":"unclaimed"}],"snapshot_sha256":"87fc3f1c0b11d0262d0fd70ebd1f21532de95a867326053d9e6009941ef316bb"},"source":{"id":"2605.13621","kind":"arxiv","version":1},"verdict":{"id":"34f44e37-e1d5-4cf8-aa8e-7cf1182cf642","model_set":{"reader":"grok-4.3"},"created_at":"2026-05-14T19:11:48.767464Z","strongest_claim":"we propose a novel detection framework WD-FQDet that explicitly decouples modality-shared and modality-specific information from infrared and visible modalities in the new view of low- and high-frequency domains, allowing fusion strategies tailored to their frequency characteristics... Experimental results on the FLIR, LLVIP, and M3FD datasets demonstrate that WD-FQDet achieves state-of-the-art performance across multiple evaluation metrics.","one_line_summary":"WD-FQDet decouples modality-shared and modality-specific features in infrared-visible images via wavelet-based frequency decomposition and frequency-aware query selection to achieve state-of-the-art detection performance.","pipeline_version":"pith-pipeline@v0.9.0","weakest_assumption":"The assumption that wavelet decomposition cleanly separates modality-shared low-frequency features from modality-specific high-frequency features and that the proposed alignment, retention, and query modules will mitigate bias and insufficiency without introducing artifacts or overfitting to the specific datasets.","pith_extraction_headline":"Wavelet decomposition decouples shared low-frequency and specific high-frequency features from infrared and visible images to improve object detection."},"references":{"count":63,"sample":[{"doi":"","year":2015,"title":"Multi-modality medical im- age fusion using discrete wavelet transform.Procedia Com- puter Science, 70:625–631, 2015","work_id":"930c14ef-2f69-426c-b7fc-48a3bdcef75a","ref_index":1,"cited_arxiv_id":"","is_internal_anchor":false},{"doi":"","year":2023,"title":"Multimodal object detection by channel switching and spatial attention","work_id":"4f218d8f-05e3-4b0e-83fc-8d0b2cd01c5c","ref_index":2,"cited_arxiv_id":"","is_internal_anchor":false},{"doi":"","year":2020,"title":"End-to- end object detection with transformers","work_id":"e5f6fe5f-1010-46d2-b656-b3f7803d7225","ref_index":3,"cited_arxiv_id":"","is_internal_anchor":false},{"doi":"","year":2023,"title":"Timothy Chase Jr, Chris Gnam, John Crassidis, and Karthik Dantu. You only crash once: Improved object detection for real-time, sim-to-real hazardous terrain detection and classi- fication for autonomo","work_id":"2d4b14db-8329-4eea-83f4-e6a240aa0e2b","ref_index":4,"cited_arxiv_id":"","is_internal_anchor":false},{"doi":"","year":2022,"title":"Multimodal object detection via probabilistic ensembling","work_id":"0a65ddb6-235a-4b26-af23-53ac568ad7d6","ref_index":5,"cited_arxiv_id":"","is_internal_anchor":false}],"resolved_work":63,"snapshot_sha256":"9dcaa972721fcf8b4f0d590f6bd7884752b172caf01805cbcfe8f242634bec0e","internal_anchors":2},"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"}