{"paper":{"title":"Pattern-Enhanced RT-DETR for Multi-Class Battery Detection","license":"http://creativecommons.org/licenses/by-nc-sa/4.0/","headline":"PaQ-RT-DETR adds pattern-based dynamic query generation to RT-DETR and raises multi-class battery detection mAP@50 to 0.782.","cross_cats":[],"primary_cat":"cs.CV","authors_text":"Enyuan Hu, Xu Zhong","submitted_at":"2026-05-13T15:29:52Z","abstract_excerpt":"Accurate and efficient battery detection is increasingly important for applications in electronic waste recycling, industrial quality control, and automated sorting systems. In this paper, we present both a comprehensive benchmark and a novel method for multi-class battery detection. We systematically compare three CNN-based detectors (YOLOv8n, YOLOv8s, YOLO11n) and two transformer-based detectors (RT-DETR-L, RT-DETR-X) on a publicly available dataset of approximately 8,591 annotated images under identical experimental conditions, and further propose PaQ-RT-DETR, which introduces pattern-based"},"claims":{"count":4,"items":[{"kind":"strongest_claim","text":"PaQ-RT-DETR-X achieves the highest overall mAP@50 of 0.782, surpassing RT-DETR-X by +2.8% with consistent per-class gains across all six battery categories including the data-scarce Bike Battery class.","source":"verdict.strongest_claim","status":"machine_extracted","claim_id":"C1","attestation":"unclaimed"},{"kind":"weakest_assumption","text":"That introducing pattern-based dynamic query generation into RT-DETR reliably alleviates query activation imbalance and produces the reported gains without hidden costs or dataset-specific artifacts.","source":"verdict.weakest_assumption","status":"machine_extracted","claim_id":"C2","attestation":"unclaimed"},{"kind":"one_line_summary","text":"PaQ-RT-DETR adds pattern-based dynamic query generation to RT-DETR, reaching 0.782 mAP@50 on 8591 battery images and beating RT-DETR-X by 2.8% across six classes.","source":"verdict.one_line_summary","status":"machine_extracted","claim_id":"C3","attestation":"unclaimed"},{"kind":"headline","text":"PaQ-RT-DETR adds pattern-based dynamic query generation to RT-DETR and raises multi-class battery detection mAP@50 to 0.782.","source":"verdict.pith_extraction.headline","status":"machine_extracted","claim_id":"C4","attestation":"unclaimed"}],"snapshot_sha256":"8d3b178754bd352ff51b0b32e50f3664ad62459302cbe3b32b39666028ad4fec"},"source":{"id":"2605.13670","kind":"arxiv","version":1},"verdict":{"id":"d0b238a3-434f-4b51-8081-be5327b7025b","model_set":{"reader":"grok-4.3"},"created_at":"2026-05-14T20:33:14.460809Z","strongest_claim":"PaQ-RT-DETR-X achieves the highest overall mAP@50 of 0.782, surpassing RT-DETR-X by +2.8% with consistent per-class gains across all six battery categories including the data-scarce Bike Battery class.","one_line_summary":"PaQ-RT-DETR adds pattern-based dynamic query generation to RT-DETR, reaching 0.782 mAP@50 on 8591 battery images and beating RT-DETR-X by 2.8% across six classes.","pipeline_version":"pith-pipeline@v0.9.0","weakest_assumption":"That introducing pattern-based dynamic query generation into RT-DETR reliably alleviates query activation imbalance and produces the reported gains without hidden costs or dataset-specific artifacts.","pith_extraction_headline":"PaQ-RT-DETR adds pattern-based dynamic query generation to RT-DETR and raises multi-class battery detection mAP@50 to 0.782."},"references":{"count":13,"sample":[{"doi":"","year":2022,"title":"Lithium-ion battery recycling—overview of techniques and trends,","work_id":"034d3ce6-be5b-4a1d-9694-51d2e34b7b50","ref_index":1,"cited_arxiv_id":"","is_internal_anchor":false},{"doi":"","year":2023,"title":"G. Jocher, A. Chaurasia, and J. Qiu, “Ultralytics YOLOv8,” https: //github.com/ultralytics/ultralytics, 2023","work_id":"87abdabc-39c4-40b2-b22f-15578f73c4ea","ref_index":2,"cited_arxiv_id":"","is_internal_anchor":false},{"doi":"","year":2024,"title":"Ultralytics, “YOLO11,” https://github.com/ultralytics/ultralytics, 2024","work_id":"a17cbbcb-a0f2-47d6-9364-648d0b5636d8","ref_index":3,"cited_arxiv_id":"","is_internal_anchor":false},{"doi":"","year":2020,"title":"End-to-end object detection with transform- ers,","work_id":"d7f21f8e-7c90-487b-a4b5-eecc18986765","ref_index":4,"cited_arxiv_id":"","is_internal_anchor":false},{"doi":"","year":2023,"title":"DINO: DETR with improved denoising anchor boxes for end-to-end object detection,","work_id":"ceb442b5-bc6f-401c-ba30-5451363536d7","ref_index":5,"cited_arxiv_id":"","is_internal_anchor":false}],"resolved_work":13,"snapshot_sha256":"f7e5ae0e182b42c333dac673fedf4e88a00c9411d6ca5c90e4de12de3b38662a","internal_anchors":0},"formal_canon":{"evidence_count":2,"snapshot_sha256":"d74848d157096f9ca2594b2ea39fc046b0cd659f4be91a952760ac6a2528c288"},"author_claims":{"count":0,"strong_count":0,"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"builder_version":"pith-number-builder-2026-05-17-v1"}