HyperAI超神経

ホームニュース最新論文チュートリアルデータセット百科事典 SOTA LLMモデル GPU ランキング学会

サイトについて

日本語

HyperAI超神経

Pose Estimation On Ochuman

評価指標

Test AP

Validation AP

評価結果

このベンチマークにおける各モデルのパフォーマンス結果

モデル名	Test AP	Validation AP	Paper Title	Repository
Associative Embedding	29.5	32.1	Associative Embedding: End-to-End Learning for Joint Detection and Grouping	-
BUCTD (CID-W32)	47.2	47.7	Rethinking pose estimation in crowds: overcoming the detection information-bottleneck and ambiguity	-
ViTPose (ViTAE-G, GT bounding boxes)	93.3	92.8	ViTPose: Simple Vision Transformer Baselines for Human Pose Estimation	-
RTMPose(RTMPose-l, GT bounding boxes)	80.3	80.5	RTMPose: Real-Time Multi-Person Pose Estimation based on MMPose	-
HQNet (ResNet-50)	40.0	-	You Only Learn One Query: Learning Unified Human Query for Single-Stage Multi-Person Multi-Task Human-Centric Perception	-
HQNet (ViT-L)	45.6	-	You Only Learn One Query: Learning Unified Human Query for Single-Stage Multi-Person Multi-Task Human-Centric Perception	-
HGG (AE+)	36.0	41.8	Differentiable Hierarchical Graph Grouping for Multi-Person Pose Estimation	-
CID (HRNet-W48)	45.0	46.1	Contextual Instance Decoupling for Robust Multi-Person Pose Estimation
ResNet-152	33.3	41.0	Simple Baselines for Human Pose Estimation and Tracking	-
ResNet-50	29.5	32.1	Simple Baselines for Human Pose Estimation and Tracking	-
BBox-Mask-Pose 2x	48.3	48.6	Detection, Pose Estimation and Segmentation for Multiple Bodies: Closing the Virtuous Circle	-
TransPose-H	-	62.3	TransPose: Keypoint Localization via Transformer	-
Associative Embedding+	32.8	40.0	Associative Embedding: End-to-End Learning for Joint Detection and Grouping	-
MIPNet (HRNet-W48)	42.5	42.0	Multi-Instance Pose Networks: Rethinking Top-Down Pose Estimation	-
RMPE	30.7	38.8	RMPE: Regional Multi-person Pose Estimation	-
MaskPose-b	45.0	45.3	Detection, Pose Estimation and Segmentation for Multiple Bodies: Closing the Virtuous Circle	-
UniHCP (direct eval)	87.4	-	UniHCP: A Unified Model for Human-Centric Perceptions	-
HRNet-W48	37.2	37.8	Multi-Instance Pose Networks: Rethinking Top-Down Pose Estimation	-

0 of 18 row(s) selected.