Multi-Person Pose Estimation on CrowdPose

Metric: mAP @0.5:0.95 (higher is better)

LeaderboardDataset

Loading chart...

Results

Hide extra data

Sort:

#	Model↕	mAP @0.5:0.95▼	Extra Data	Paper	Date↕	Code
1	RTMO-l	83.8	Yes	RTMO: Towards High-Performance One-Stage Real-Ti...	2023-12-12	Code
2	BUCTD-W48 (w/cond. input from PETR, and generative sampling)	78.5	No	Rethinking pose estimation in crowds: overcoming...	2023-06-13	Code
3	I²R-Net (1st stage: HRFormer-B)	77.4	No	I^2R-Net: Intra- and Inter-Human Relation Networ...	2022-06-22	Code
4	ED-Pose (Swin-L)	76.6	Yes	Explicit Box Detection Unifies End-to-End Multi-...	2023-02-03	Code
5	DETRPose-X	75.1	No	DETRPose: Real-time end-to-end transformer model...	2025-06-16	Code
6	DETRPose-L	73.3	No	DETRPose: Real-time end-to-end transformer model...	2025-06-16	Code
7	HRFormer-B	72.4	No	HRFormer: High-Resolution Transformer for Dense ...	2021-10-18	Code
8	BAPose (W32)	72.2	No	BAPose: Bottom-Up Pose Estimation with Disentang...	2021-12-20	Code
9	DETRPose-M	72	No	DETRPose: Real-time end-to-end transformer model...	2025-06-16	Code
10	TransPose-H	71.8	No	TransPose: Keypoint Localization via Transformer	2020-12-28	Code
11	SCIO (HRNet-48)	71.5	No	Self-Constrained Inference Optimization on Struc...	2022-07-06	-
12	HigherHRNet (ScaleNet_P4)	71.3	Yes	ScaleNAS: One-Shot Learning of Scale-Aware Repre...	2020-11-30	-
13	MIPNet (HRNet-W48)	70	No	Multi-Instance Pose Networks: Rethinking Top-Dow...	2021-01-27	Code
14	CenterGroup	69.4	No	The Center of Attention: Center-Keypoint Groupin...	2021-10-11	Code
15	HigherHRNet(HR-Net-48)	67.6	Yes	HigherHRNet: Scale-Aware Representation Learning...	2019-08-27	Code
16	DETRPose-S	67.4	No	DETRPose: Real-time end-to-end transformer model...	2025-06-16	Code
17	Joint-candidate SPPE +	66	No	CrowdPose: Efficient Crowded Scenes Pose Estimat...	2018-12-02	Code
18	OccNet	65.5	No	Human Pose Estimation for Real-World Crowded Sce...	2019-07-16	Code
19	Hourglass-104	65.2	No	Greedy Offset-Guided Keypoint Grouping for Human...	2021-07-07	Code
20	SPM	63.7	No	Single-Stage Multi-Person Pose Machines	2019-08-24	Code
21	AlphaPose	61	No	RMPE: Regional Multi-person Pose Estimation	2016-12-01	Code
22	Simple baseline	60.8	No	Simple Baselines for Human Pose Estimation and T...	2018-04-17	Code
23	ROMP+CAR	58.6	No	Monocular, One-stage, Regression of Multiple 3D ...	2020-08-27	Code
24	LitePose-S	58.3	No	Lite Pose: Efficient Architecture Design for 2D ...	2022-05-03	Code
25	Mask R-CNN	57.2	No	Mask R-CNN	2017-03-20	Code
26	DETRPose-N	56	No	DETRPose: Real-time end-to-end transformer model...	2025-06-16	Code
27	ROMP	55.6	No	Monocular, One-stage, Regression of Multiple 3D ...	2020-08-27	Code

#1RTMO-lSOTA
83.8
mAP @0.5:0.95· Extra Data· 2023-12-12
RTMO: Towards High-Performance One-Stage Real-Time Multi-Person Pose Estimation Code
#2BUCTD-W48 (w/cond. input from PETR, and generative sampling)SOTA
78.5
mAP @0.5:0.95· 2023-06-13
Rethinking pose estimation in crowds: overcoming the detection information-bottleneck and ambiguity Code
#3I²R-Net (1st stage: HRFormer-B)SOTA
77.4
mAP @0.5:0.95· 2022-06-22
I^2R-Net: Intra- and Inter-Human Relation Network for Multi-Person Pose Estimation Code
#4ED-Pose (Swin-L)
76.6
mAP @0.5:0.95· Extra Data· 2023-02-03
Explicit Box Detection Unifies End-to-End Multi-Person Pose Estimation Code
#5DETRPose-X
75.1
mAP @0.5:0.95· 2025-06-16
DETRPose: Real-time end-to-end transformer model for multi-person pose estimation Code
#6DETRPose-L
73.3
mAP @0.5:0.95· 2025-06-16
DETRPose: Real-time end-to-end transformer model for multi-person pose estimation Code
#7HRFormer-BSOTA
72.4
mAP @0.5:0.95· 2021-10-18
HRFormer: High-Resolution Transformer for Dense Prediction Code
#8BAPose (W32)
72.2
mAP @0.5:0.95· 2021-12-20
BAPose: Bottom-Up Pose Estimation with Disentangled Waterfall Representations Code
#9DETRPose-M
72
mAP @0.5:0.95· 2025-06-16
DETRPose: Real-time end-to-end transformer model for multi-person pose estimation Code
#10TransPose-HSOTA
71.8
mAP @0.5:0.95· 2020-12-28
TransPose: Keypoint Localization via Transformer Code
#11SCIO (HRNet-48)
71.5
mAP @0.5:0.95· 2022-07-06
Self-Constrained Inference Optimization on Structural Groups for Human Pose Estimation
#12HigherHRNet (ScaleNet_P4)SOTA
71.3
mAP @0.5:0.95· Extra Data· 2020-11-30
ScaleNAS: One-Shot Learning of Scale-Aware Representations for Visual Recognition
#13MIPNet (HRNet-W48)
70
mAP @0.5:0.95· 2021-01-27
Multi-Instance Pose Networks: Rethinking Top-Down Pose Estimation Code
#14CenterGroup
69.4
mAP @0.5:0.95· 2021-10-11
The Center of Attention: Center-Keypoint Grouping via Attention for Multi-Person Pose Estimation Code
#15HigherHRNet(HR-Net-48)SOTA
67.6
mAP @0.5:0.95· Extra Data· 2019-08-27
HigherHRNet: Scale-Aware Representation Learning for Bottom-Up Human Pose Estimation Code
#16DETRPose-S
67.4
mAP @0.5:0.95· 2025-06-16
DETRPose: Real-time end-to-end transformer model for multi-person pose estimation Code
#17Joint-candidate SPPE +SOTA
66
mAP @0.5:0.95· 2018-12-02
CrowdPose: Efficient Crowded Scenes Pose Estimation and A New Benchmark Code
#18OccNet
65.5
mAP @0.5:0.95· 2019-07-16
Human Pose Estimation for Real-World Crowded Scenarios Code
#19Hourglass-104
65.2
mAP @0.5:0.95· 2021-07-07
Greedy Offset-Guided Keypoint Grouping for Human Pose Estimation Code
#20SPM
63.7
mAP @0.5:0.95· 2019-08-24
Single-Stage Multi-Person Pose Machines Code
#21AlphaPoseSOTA
61
mAP @0.5:0.95· 2016-12-01
RMPE: Regional Multi-person Pose Estimation Code
#22Simple baseline
60.8
mAP @0.5:0.95· 2018-04-17
Simple Baselines for Human Pose Estimation and Tracking Code
#23ROMP+CAR
58.6
mAP @0.5:0.95· 2020-08-27
Monocular, One-stage, Regression of Multiple 3D People Code
#24LitePose-S
58.3
mAP @0.5:0.95· 2022-05-03
Lite Pose: Efficient Architecture Design for 2D Human Pose Estimation Code
#25Mask R-CNN
57.2
mAP @0.5:0.95· 2017-03-20
Mask R-CNN Code
#26DETRPose-N
56
mAP @0.5:0.95· 2025-06-16
DETRPose: Real-time end-to-end transformer model for multi-person pose estimation Code
#27ROMP
55.6
mAP @0.5:0.95· 2020-08-27
Monocular, One-stage, Regression of Multiple 3D People Code