Papers › DeBiFormer: Vision Transformer with Deformable Agent Bi-level Routing Attention

DeBiFormer: Vision Transformer with Deformable Agent Bi-level Routing Attention

11 Oct 2024arXiv:2410.08582archive 2025-07-28

Nguyen Huu Bao Long, Chenyu Zhang, Yuzhi Shi, Tsubasa Hirakawa, Takayoshi Yamashita, Tohgoroh Matsui, Hironobu Fujiyoshi

Vision Transformers with various attention modules have demonstrated superior performance on vision tasks. While using sparsity-adaptive attention, such as in DAT, has yielded strong results in image classification, the key-value pairs selected by deformable points lack semantic relevance when fine-tuning for semantic segmentation tasks. The query-aware sparsity attention in BiFormer seeks to focus each query on top-k routed regions. However, during attention calculation, the selected key-value pairs are influenced by too many irrelevant queries, reducing attention on the more important ones. To address these issues, we propose the Deformable Bi-level Routing Attention (DBRA) module, which optimizes the selection of key-value pairs using agent queries and enhances the interpretability of queries in attention maps. Based on this, we introduce the Deformable Bi-level Routing Attention Transformer (DeBiFormer), a novel general-purpose vision transformer built with the DBRA module. DeBiFormer has been validated on various computer vision tasks, including image classification, object detection, and semantic segmentation, providing strong evidence of its effectiveness.Code is available at {https://github.com/maclong01/DeBiFormer}

PaperPDFCode

In Syntology Open this paper in Syntology's Atlas, the map of the papers in Syntology's graph and their citations.

Code

maclong01/DeBiFormer officialmentioned in papermentioned on GitHubpytorch report

Repository list and official/mentioned flags are the archive's, frozen 2025-07-28. Reachability, where shown, is from one Syntology probe window (2026-09-16 to 2026-09-18); repositories not probed show nothing. GitHub stars are not tracked.

Code Syntology ran Syntology

Not run by Syntology. Nothing on this page verifies that the listed code works.

Tasks

Image ClassificationObject DetectionSemantic Segmentationimage-classificationobject-detection

Results from the paper archive 2025-07-28

TaskDatasetModelMetricValueRank at snapshotLeaderboardReport
Image Classification ImageNet DeBiFormer-B Top 1 Accuracy 84.4% #317 of 1060 Archive leaderboard report
Image Classification ImageNet DeBiFormer-S Top 1 Accuracy 83.9% #375 of 1060 Archive leaderboard report
Image Classification ImageNet DeBiFormer-T Top 1 Accuracy 81.9% #596 of 1060 Archive leaderboard report
Object Detection COCO 2017 DeBiFormer-B (IN1k pretrain, MaskRCNN 12ep) mAP 48.5 #17 of 24 Archive leaderboard report
Object Detection COCO 2017 DeBiFormer-S (IN1k pretrain, MaskRCNN 12ep) mAP 47.5 #19 of 24 Archive leaderboard report
Object Detection COCO 2017 DeBiFormer-B (IN1k pretrain, Retina) mAP 47.1 #20 of 24 Archive leaderboard report
Object Detection COCO 2017 DeBiFormer-S (IN1k pretrain, Retina) mAP 45.6 #21 of 24 Archive leaderboard report
Semantic Segmentation ADE20K DeBiFormer-B (IN1k pretrain, Upernet 160k) Validation mIoU 52.0 #90 of 235 Archive leaderboard report

Ranks are positions in the archive's leaderboards as they stood at the 2025-07-28 snapshot. Results published since then are not among these rows, so a rank here is not a current standing.

Methods

Absolute Position EncodingsAdamAttentionBPEDense ConnectionsDropoutFocusLabel SmoothingLayer NormalizationLinear LayerMulti-Head AttentionPosition-Wise Feed-Forward LayerResidual ConnectionRouting AttentionSoftmaxTransformerVision Transformer

Report a problem or propose a change · a person checks every report against the paper or source before anything changes; decisions are listed on /corrections