Papers › Anchor DETR: Query Design for Transformer-Based Object Detection

Anchor DETR: Query Design for Transformer-Based Object Detection

15 Sep 2021arXiv:2109.07107archive 2025-07-28

Yingming Wang, Xiangyu Zhang, Tong Yang, Jian Sun

In this paper, we propose a novel query design for the transformer-based object detection. In previous transformer-based detectors, the object queries are a set of learned embeddings. However, each learned embedding does not have an explicit physical meaning and we cannot explain where it will focus on. It is difficult to optimize as the prediction slot of each object query does not have a specific mode. In other words, each object query will not focus on a specific region. To solved these problems, in our query design, object queries are based on anchor points, which are widely used in CNN-based detectors. So each object query focuses on the objects near the anchor point. Moreover, our query design can predict multiple objects at one position to solve the difficulty: "one region, multiple objects". In addition, we design an attention variant, which can reduce the memory cost while achieving similar or better performance than the standard attention in DETR. Thanks to the query design and the attention variant, the proposed detector that we called Anchor DETR, can achieve better performance and run faster than the DETR with 10× fewer training epochs. For example, it achieves 44.2 AP with 19 FPS on the MSCOCO dataset when using the ResNet50-DC5 feature for training 50 epochs. Extensive experiments on the MSCOCO benchmark prove the effectiveness of the proposed methods. Code is available at \url{https://github.com/megvii-research/AnchorDETR}.

PaperPDFCode

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

Code

megvii-model/anchordetr officialmentioned in papermentioned on GitHubpytorchNOASSERTION report
megvii-research/AnchorDETR officialmentioned in papermentioned on GitHubpytorchNOASSERTION 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

ObjectObject Detectionobject-detection

Results from the paper archive 2025-07-28

TaskDatasetModelMetricValueRank at snapshotLeaderboardReport
Object Detection COCO minival Anchor DETR-DC5-R101 AP50 65.7 #115 of 220 Archive leaderboard report
Object Detection COCO minival Anchor DETR-DC5-R101 AP75 48.8 #115 of 220 Archive leaderboard report
Object Detection COCO minival Anchor DETR-DC5-R101 APL 61.6 #115 of 220 Archive leaderboard report
Object Detection COCO minival Anchor DETR-DC5-R101 APM 49.4 #115 of 220 Archive leaderboard report
Object Detection COCO minival Anchor DETR-DC5-R101 APS 25.8 #115 of 220 Archive leaderboard report
Object Detection COCO minival Anchor DETR-DC5-R101 box AP 45.1 #115 of 220 Archive leaderboard report
Object Detection COCO minival Anchor DETR-DC5-R50 AP50 64.7 #134 of 220 Archive leaderboard report
Object Detection COCO minival Anchor DETR-DC5-R50 AP75 47.5 #134 of 220 Archive leaderboard report
Object Detection COCO minival Anchor DETR-DC5-R50 APL 60.6 #134 of 220 Archive leaderboard report
Object Detection COCO minival Anchor DETR-DC5-R50 APM 48.2 #134 of 220 Archive leaderboard report
Object Detection COCO minival Anchor DETR-DC5-R50 APS 24.7 #134 of 220 Archive leaderboard report
Object Detection COCO minival Anchor DETR-DC5-R50 box AP 44.2 #134 of 220 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 EncodingsAdamAttentionBPEConvolutionDense ConnectionsDetrDropoutFeedforward NetworkLabel SmoothingLayer NormalizationLinear LayerMulti-Head AttentionPosition-Wise Feed-Forward LayerResidual ConnectionSoftmaxTransformer

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