3D on COCO test-dev

Metric: AP (higher is better)

LeaderboardDataset

Loading chart...

Results

Hide augmentations

Sort:

#	Model↕	AP▼	Augmentations	Paper	Date↕	Code
1	ViTPose (ViTAE-G, ensemble)	81.1	No	ViTPose: Simple Vision Transformer Baselines for...	2022-04-26	Code
2	ViTPose (ViTAE-G)	80.9	No	ViTPose: Simple Vision Transformer Baselines for...	2022-04-26	Code
3	UDP-Pose-PSA(384x288)	79.5	No	Polarized Self-Attention: Towards High-quality P...	2021-07-02	Code
4	PoseBH-H	79.5	Yes	PoseBH: Prototypical Multi-Dataset Training Beyo...	2025-05-23	Code
5	4xRSN-50 (ensemble)	79.2	No	Learning Delicate Local Representations for Mult...	2020-03-09	Code
6	SCIO (HRNet-48)	79.2	No	Self-Constrained Inference Optimization on Struc...	2022-07-06	-
7	CCM+	78.9	No	Towards High Performance Human Keypoint Detection	2020-02-03	Code
8	UDP-Pose-PSA(256x192)	78.9	No	Polarized Self-Attention: Towards High-quality P...	2021-07-02	Code
9	HRNet-W48plus	78.7	No	AID: Pushing the Performance Boundary of Human P...	2020-08-17	Code
10	4xRSN-50	78.6	No	Learning Delicate Local Representations for Mult...	2020-03-09	Code
11	PCT (256x256)	78.3	No	Human Pose as Compositional Tokens	2023-03-21	Code
12	HRNet-W48+DARK	77.4	No	Distribution-Aware Coordinate Representation for...	2019-10-14	Code
13	SwinV2-L 1K-MIM	77.2	No	Revealing the Dark Secrets of Masked Image Model...	2022-05-26	Code
14	HRNet-W48 + extra data	77	No	Deep High-Resolution Representation Learning for...	2019-02-25	Code
15	EvoPose2D-L	76.8	No	EvoPose2D: Pushing the Boundaries of 2D Human Po...	2020-11-17	Code
16	SwinV2-B 1K-MIM	76.7	No	Revealing the Dark Secrets of Masked Image Model...	2022-05-26	Code
17	HRNet-W48+UDP	76.5	No	The Devil is in the Details: Delving into Unbias...	2019-11-18	Code
18	OmniPose (WASPv2)	76.4	No	OmniPose: A Multi-Scale Framework for Multi-Pers...	2021-03-18	Code
19	HRFormer-B	76.2	No	HRFormer: High-Resolution Transformer for Dense ...	2021-10-18	Code
20	HRNet-W32	76.2	No	AID: Pushing the Performance Boundary of Human P...	2020-08-17	Code
21	MSPN	76.1	No	Rethinking on Multi-Stage Networks for Human Pos...	2019-01-01	Code
22	MSPN	76.1	No	Rethinking on Multi-Stage Networks for Human Pos...	2019-01-01	Code
23	MIPNet	75.7	No	Multi-Instance Pose Networks: Rethinking Top-Dow...	2021-01-27	Code
24	PPE (ResNeXt-101)	75.7	No	Deep Multi-Task Networks For Occluded Pedestrian...	2022-06-15	-
25	TransPose-H-A6	75	No	TransPose: Keypoint Localization via Transformer	2020-12-28	Code
26	PoseFix	74.7	No	PoseFix: Model-agnostic General Human Pose Refin...	2018-12-10	Code
27	DPIT-L	74.6	No	DPIT: Dual-Pipeline Integrated Transformer for H...	2022-09-02	-
28	S-ViPNAS-HRNetW32	73.9	No	ViPNAS: Efficient Video Pose Estimation via Neur...	2021-05-21	Code
29	Flow-based (ResNet-152)	73.7	No	Simple Baselines for Human Pose Estimation and T...	2018-04-17	Code
30	ResNet50	73.7	No	AID: Pushing the Performance Boundary of Human P...	2020-08-17	Code
31	CPN+ [6, 9]	73	No	Cascaded Pyramid Network for Multi-Person Pose E...	2017-11-20	Code
32	RMPE++	72.3	No	RMPE: Regional Multi-person Pose Estimation	2016-12-01	Code
33	TFPose (ND=6 ResNet-50)	72.2	No	TFPose: Direct Human Pose Estimation with Transf...	2021-03-29	-
34	CPN	72.1	No	Cascaded Pyramid Network for Multi-Person Pose E...	2017-11-20	Code
35	HigherHRNet (ScaleNet_P4)	71.6	No	ScaleNAS: One-Shot Learning of Scale-Aware Repre...	2020-11-30	-
36	OpenPifPaf	70.9	No	OpenPifPaf: Composite Fields for Semantic Keypoi...	2021-03-03	Code
37	LOGO-CAP (Ours) HRNet-W48	70.8	No	-	-	-
38	Dite-HRNet-30	70.6	No	Dite-HRNet: Dynamic Lightweight High-Resolution ...	2022-04-22	Code
39	HigherHRNet (HR-Net-48)	70.5	No	HigherHRNet: Scale-Aware Representation Learning...	2019-08-27	Code
40	KAPAO-L	70.3	No	Rethinking Keypoint Representations: Modeling Ke...	2021-11-16	Code
41	S-ViPNAS-Res50	70.3	No	ViPNAS: Efficient Video Pose Estimation via Neur...	2021-05-21	Code
42	SMPR (HR-Net-32)	70.2	No	SMPR: Single-Stage Multi-Person Pose Regression	2020-06-28	Code
43	Lite-HRNet-30	69.7	Yes	Lite-HRNet: A Lightweight High-Resolution Network	2021-04-13	Code
44	KAPAO-M	68.8	No	Rethinking Keypoint Representations: Modeling Ke...	2021-11-16	Code
45	PersonLab	68.7	No	PersonLab: Person Pose Estimation and Instance S...	2018-03-22	Code
46	Simple Pose	68.1	Yes	Simple Pose: Rethinking and Improving a Bottom-u...	2019-11-24	Code
47	Simple Pose	68.1	No	Simple Pose: Rethinking and Improving a Bottom-u...	2019-11-24	Code
48	Identity Mapping Hourglass	68.1	No	Simple Pose: Rethinking and Improving a Bottom-u...	2019-11-24	Code
49	Lite-HRNet-18	66.9	No	Lite-HRNet: A Lightweight High-Resolution Network	2021-04-13	Code
50	SPM	66.9	No	Single-Stage Multi-Person Pose Machines	2019-08-24	Code
51	Supervising Self-Attention	66.5	No	Attend to Who You Are: Supervising Self-Attentio...	2021-11-25	Code
52	PifPaf (single-scale)	66.4	No	PifPaf: Composite Fields for Human Pose Estimation	2019-03-15	Code
53	Hourglass-104	65.6	No	Greedy Offset-Guided Keypoint Grouping for Human...	2021-07-07	Code
54	G-RMI	64.9	No	Towards Accurate Multi-person Pose Estimation in...	2017-01-06	-
55	G-RMI	64.9	No	Towards Accurate Multi-person Pose Estimation in...	2017-01-06	-
56	DirectPose (ResNet-101)	64.8	No	DirectPose: Direct End-to-End Multi-Person Pose ...	2019-11-18	Code
57	Faster R-CNN (ImageNet+300M)	64.4	Yes	Revisiting Unreasonable Effectiveness of Data in...	2017-07-10	Code
58	OpenPose	64.2	Yes	OpenPose: Realtime Multi-Person 2D Pose Estimati...	2018-12-18	Code
59	KAPAO-S	63.8	No	Rethinking Keypoint Representations: Modeling Ke...	2021-11-16	Code
60	DirectPose (ResNet-101)	63.3	No	DirectPose: Direct End-to-End Multi-Person Pose ...	2019-11-18	Code
61	Mask-RCNN	63.1	No	Mask R-CNN	2017-03-20	Code
62	CMU-Pose	61.8	No	Realtime Multi-Person 2D Pose Estimation using P...	2016-11-24	Code
63	RMPE	61.8	No	RMPE: Regional Multi-person Pose Estimation	2016-12-01	Code
64	CMU-Pose	61.8	No	Realtime Multi-Person 2D Pose Estimation using P...	2016-11-24	Code
65	RMPE	61.8	No	RMPE: Regional Multi-person Pose Estimation	2016-12-01	Code

#1ViTPose (ViTAE-G, ensemble)SOTA
81.1
AP· 2022-04-26
ViTPose: Simple Vision Transformer Baselines for Human Pose Estimation Code
#2ViTPose (ViTAE-G)
80.9
AP· 2022-04-26
ViTPose: Simple Vision Transformer Baselines for Human Pose Estimation Code
#3UDP-Pose-PSA(384x288)SOTA
79.5
AP· 2021-07-02
Polarized Self-Attention: Towards High-quality Pixel-wise Regression Code
#4PoseBH-H
79.5
AP· Augmentations· 2025-05-23
PoseBH: Prototypical Multi-Dataset Training Beyond Human Pose Estimation Code
#54xRSN-50 (ensemble)SOTA
79.2
AP· 2020-03-09
Learning Delicate Local Representations for Multi-Person Pose Estimation Code
#6SCIO (HRNet-48)
79.2
AP· 2022-07-06
Self-Constrained Inference Optimization on Structural Groups for Human Pose Estimation
#7CCM+SOTA
78.9
AP· 2020-02-03
Towards High Performance Human Keypoint Detection Code
#8UDP-Pose-PSA(256x192)
78.9
AP· 2021-07-02
Polarized Self-Attention: Towards High-quality Pixel-wise Regression Code
#9HRNet-W48plus
78.7
AP· 2020-08-17
AID: Pushing the Performance Boundary of Human Pose Estimation with Information Dropping Augmentation Code
#104xRSN-50
78.6
AP· 2020-03-09
Learning Delicate Local Representations for Multi-Person Pose Estimation Code
#11PCT (256x256)
78.3
AP· 2023-03-21
Human Pose as Compositional Tokens Code
#12HRNet-W48+DARKSOTA
77.4
AP· 2019-10-14
Distribution-Aware Coordinate Representation for Human Pose Estimation Code
#13SwinV2-L 1K-MIM
77.2
AP· 2022-05-26
Revealing the Dark Secrets of Masked Image Modeling Code
#14HRNet-W48 + extra dataSOTA
77
AP· 2019-02-25
Deep High-Resolution Representation Learning for Human Pose Estimation Code
#15EvoPose2D-L
76.8
AP· 2020-11-17
EvoPose2D: Pushing the Boundaries of 2D Human Pose Estimation using Accelerated Neuroevolution with Weight Transfer Code
#16SwinV2-B 1K-MIM
76.7
AP· 2022-05-26
Revealing the Dark Secrets of Masked Image Modeling Code
#17HRNet-W48+UDP
76.5
AP· 2019-11-18
The Devil is in the Details: Delving into Unbiased Data Processing for Human Pose Estimation Code
#18OmniPose (WASPv2)
76.4
AP· 2021-03-18
OmniPose: A Multi-Scale Framework for Multi-Person Pose Estimation Code
#19HRFormer-B
76.2
AP· 2021-10-18
HRFormer: High-Resolution Transformer for Dense Prediction Code
#20HRNet-W32
76.2
AP· 2020-08-17
AID: Pushing the Performance Boundary of Human Pose Estimation with Information Dropping Augmentation Code
#21MSPNSOTA
76.1
AP· 2019-01-01
Rethinking on Multi-Stage Networks for Human Pose Estimation Code
#22MSPN
76.1
AP· 2019-01-01
Rethinking on Multi-Stage Networks for Human Pose Estimation Code
#23MIPNet
75.7
AP· 2021-01-27
Multi-Instance Pose Networks: Rethinking Top-Down Pose Estimation Code
#24PPE (ResNeXt-101)
75.7
AP· 2022-06-15
Deep Multi-Task Networks For Occluded Pedestrian Pose Estimation
#25TransPose-H-A6
75
AP· 2020-12-28
TransPose: Keypoint Localization via Transformer Code
#26PoseFixSOTA
74.7
AP· 2018-12-10
PoseFix: Model-agnostic General Human Pose Refinement Network Code
#27DPIT-L
74.6
AP· 2022-09-02
DPIT: Dual-Pipeline Integrated Transformer for Human Pose Estimation
#28S-ViPNAS-HRNetW32
73.9
AP· 2021-05-21
ViPNAS: Efficient Video Pose Estimation via Neural Architecture Search Code
#29Flow-based (ResNet-152)SOTA
73.7
AP· 2018-04-17
Simple Baselines for Human Pose Estimation and Tracking Code
#30ResNet50
73.7
AP· 2020-08-17
AID: Pushing the Performance Boundary of Human Pose Estimation with Information Dropping Augmentation Code
#31CPN+ [6, 9]SOTA
73
AP· 2017-11-20
Cascaded Pyramid Network for Multi-Person Pose Estimation Code
#32RMPE++SOTA
72.3
AP· 2016-12-01
RMPE: Regional Multi-person Pose Estimation Code
#33TFPose (ND=6 ResNet-50)
72.2
AP· 2021-03-29
TFPose: Direct Human Pose Estimation with Transformers
#34CPN
72.1
AP· 2017-11-20
Cascaded Pyramid Network for Multi-Person Pose Estimation Code
#35HigherHRNet (ScaleNet_P4)
71.6
AP· 2020-11-30
ScaleNAS: One-Shot Learning of Scale-Aware Representations for Visual Recognition
#36OpenPifPaf
70.9
AP· 2021-03-03
OpenPifPaf: Composite Fields for Semantic Keypoint Detection and Spatio-Temporal Association Code
#37LOGO-CAP (Ours) HRNet-W48
70.8
AP
No paper
#38Dite-HRNet-30
70.6
AP· 2022-04-22
Dite-HRNet: Dynamic Lightweight High-Resolution Network for Human Pose Estimation Code
#39HigherHRNet (HR-Net-48)
70.5
AP· 2019-08-27
HigherHRNet: Scale-Aware Representation Learning for Bottom-Up Human Pose Estimation Code
#40KAPAO-L
70.3
AP· 2021-11-16
Rethinking Keypoint Representations: Modeling Keypoints and Poses as Objects for Multi-Person Human Pose Estimation Code
#41S-ViPNAS-Res50
70.3
AP· 2021-05-21
ViPNAS: Efficient Video Pose Estimation via Neural Architecture Search Code
#42SMPR (HR-Net-32)
70.2
AP· 2020-06-28
SMPR: Single-Stage Multi-Person Pose Regression Code
#43Lite-HRNet-30
69.7
AP· Augmentations· 2021-04-13
Lite-HRNet: A Lightweight High-Resolution Network Code
#44KAPAO-M
68.8
AP· 2021-11-16
Rethinking Keypoint Representations: Modeling Keypoints and Poses as Objects for Multi-Person Human Pose Estimation Code
#45PersonLab
68.7
AP· 2018-03-22
PersonLab: Person Pose Estimation and Instance Segmentation with a Bottom-Up, Part-Based, Geometric Embedding Model Code
#46Simple Pose
68.1
AP· Augmentations· 2019-11-24
Simple Pose: Rethinking and Improving a Bottom-up Approach for Multi-Person Pose Estimation Code
#47Simple Pose
68.1
AP· 2019-11-24
Simple Pose: Rethinking and Improving a Bottom-up Approach for Multi-Person Pose Estimation Code
#48Identity Mapping Hourglass
68.1
AP· 2019-11-24
Simple Pose: Rethinking and Improving a Bottom-up Approach for Multi-Person Pose Estimation Code
#49Lite-HRNet-18
66.9
AP· 2021-04-13
Lite-HRNet: A Lightweight High-Resolution Network Code
#50SPM
66.9
AP· 2019-08-24
Single-Stage Multi-Person Pose Machines Code
#51Supervising Self-Attention
66.5
AP· 2021-11-25
Attend to Who You Are: Supervising Self-Attention for Keypoint Detection and Instance-Aware Association Code
#52PifPaf (single-scale)
66.4
AP· 2019-03-15
PifPaf: Composite Fields for Human Pose Estimation Code
#53Hourglass-104
65.6
AP· 2021-07-07
Greedy Offset-Guided Keypoint Grouping for Human Pose Estimation Code
#54G-RMI
64.9
AP· 2017-01-06
Towards Accurate Multi-person Pose Estimation in the Wild
#55G-RMI
64.9
AP· 2017-01-06
Towards Accurate Multi-person Pose Estimation in the Wild
#56DirectPose (ResNet-101)
64.8
AP· 2019-11-18
DirectPose: Direct End-to-End Multi-Person Pose Estimation Code
#57Faster R-CNN (ImageNet+300M)
64.4
AP· Augmentations· 2017-07-10
Revisiting Unreasonable Effectiveness of Data in Deep Learning Era Code
#58OpenPose
64.2
AP· Augmentations· 2018-12-18
OpenPose: Realtime Multi-Person 2D Pose Estimation using Part Affinity Fields Code
#59KAPAO-S
63.8
AP· 2021-11-16
Rethinking Keypoint Representations: Modeling Keypoints and Poses as Objects for Multi-Person Human Pose Estimation Code
#60DirectPose (ResNet-101)
63.3
AP· 2019-11-18
DirectPose: Direct End-to-End Multi-Person Pose Estimation Code
#61Mask-RCNN
63.1
AP· 2017-03-20
Mask R-CNN Code
#62CMU-PoseSOTA
61.8
AP· 2016-11-24
Realtime Multi-Person 2D Pose Estimation using Part Affinity Fields Code
#63RMPE
61.8
AP· 2016-12-01
RMPE: Regional Multi-person Pose Estimation Code
#64CMU-Pose
61.8
AP· 2016-11-24
Realtime Multi-Person 2D Pose Estimation using Part Affinity Fields Code
#65RMPE
61.8
AP· 2016-12-01
RMPE: Regional Multi-person Pose Estimation Code