16k on COCO 2017 val

Metric: AP50 (higher is better)

LeaderboardDataset

Loading chart...

Results

Hide augmentations

Sort:

#	Model↕	AP50▼	Augmentations	Paper	Date↕	Code
1	Mr. DETR (Swin-L, 1x, 5cale)	79	Yes	Mr. DETR: Instructive Multi-Route Training for D...	2024-12-13	Code
2	MI-DETR (Swin-L 1x)	76.5	No	MI-DETR: An Object Detection Model with Multi-ti...	2025-03-03	Code
3	Relation-DETR (Swin-L 2x)	76.4	No	Relation DETR: Exploring Explicit Position Relat...	2024-07-16	Code
4	Mr. DETR (Swin-L, 1x, 4scale)	76.3	No	Mr. DETR: Instructive Multi-Route Training for D...	2024-12-13	Code
5	Relation-DETR (Swin-L 1x)	76.1	No	Relation DETR: Exploring Explicit Position Relat...	2024-07-16	Code
6	Salience-DETR (Focal-L 1x)	75.5	No	Salience DETR: Enhancing Detection Transformer w...	2024-03-24	Code
7	Salience-DETR (Swin-L 1x)	75	No	Salience DETR: Enhancing Detection Transformer w...	2024-03-24	Code
8	YOLOv6-L6(46 fps, V100, bs1)	74.5	No	YOLOv6 v3.0: A Full-Scale Reloading	2023-01-13	Code
9	Relation-DETR (ResNet50 2x)	69.7	No	Relation DETR: Exploring Explicit Position Relat...	2024-07-16	Code
10	ViDT Swin-base	69.4	No	ViDT: An Efficient and Effective Fully Transform...	2021-10-08	Code
11	Relation-DETR (ResNet50 1x)	69.1	No	Relation DETR: Exploring Explicit Position Relat...	2024-07-16	Code
12	DyHead (Swin-T, multi scale)	68	No	Dynamic Head: Unifying Object Detection Heads wi...	2021-06-15	Code
13	Salience-DETR (ResNet50 1x)	67.7	No	Salience DETR: Enhancing Detection Transformer w...	2024-03-24	Code
14	ViDT Swin-small	67.7	No	ViDT: An Efficient and Effective Fully Transform...	2021-10-08	Code
15	ViDT Swin-tiny	64.5	No	ViDT: An Efficient and Effective Fully Transform...	2021-10-08	Code
16	ViDT Swin-nano	59.6	No	ViDT: An Efficient and Effective Fully Transform...	2021-10-08	Code

#1Mr. DETR (Swin-L, 1x, 5cale)SOTA
79
AP50· Augmentations· 2024-12-13
Mr. DETR: Instructive Multi-Route Training for Detection Transformers Code
#2MI-DETR (Swin-L 1x)
76.5
AP50· 2025-03-03
MI-DETR: An Object Detection Model with Multi-time Inquiries Mechanism Code
#3Relation-DETR (Swin-L 2x)SOTA
76.4
AP50· 2024-07-16
Relation DETR: Exploring Explicit Position Relation Prior for Object Detection Code
#4Mr. DETR (Swin-L, 1x, 4scale)
76.3
AP50· 2024-12-13
Mr. DETR: Instructive Multi-Route Training for Detection Transformers Code
#5Relation-DETR (Swin-L 1x)
76.1
AP50· 2024-07-16
Relation DETR: Exploring Explicit Position Relation Prior for Object Detection Code
#6Salience-DETR (Focal-L 1x)SOTA
75.5
AP50· 2024-03-24
Salience DETR: Enhancing Detection Transformer with Hierarchical Salience Filtering Refinement Code
#7Salience-DETR (Swin-L 1x)
75
AP50· 2024-03-24
Salience DETR: Enhancing Detection Transformer with Hierarchical Salience Filtering Refinement Code
#8YOLOv6-L6(46 fps, V100, bs1)SOTA
74.5
AP50· 2023-01-13
YOLOv6 v3.0: A Full-Scale Reloading Code
#9Relation-DETR (ResNet50 2x)
69.7
AP50· 2024-07-16
Relation DETR: Exploring Explicit Position Relation Prior for Object Detection Code
#10ViDT Swin-baseSOTA
69.4
AP50· 2021-10-08
ViDT: An Efficient and Effective Fully Transformer-based Object Detector Code
#11Relation-DETR (ResNet50 1x)
69.1
AP50· 2024-07-16
Relation DETR: Exploring Explicit Position Relation Prior for Object Detection Code
#12DyHead (Swin-T, multi scale)SOTA
68
AP50· 2021-06-15
Dynamic Head: Unifying Object Detection Heads with Attentions Code
#13Salience-DETR (ResNet50 1x)
67.7
AP50· 2024-03-24
Salience DETR: Enhancing Detection Transformer with Hierarchical Salience Filtering Refinement Code
#14ViDT Swin-small
67.7
AP50· 2021-10-08
ViDT: An Efficient and Effective Fully Transformer-based Object Detector Code
#15ViDT Swin-tiny
64.5
AP50· 2021-10-08
ViDT: An Efficient and Effective Fully Transformer-based Object Detector Code
#16ViDT Swin-nano
59.6
AP50· 2021-10-08
ViDT: An Efficient and Effective Fully Transformer-based Object Detector Code