HyperAI초신경

홈 뉴스 최신 연구 논문 튜토리얼 데이터셋 백과사전 SOTA LLM 모델 GPU 랭킹 컨퍼런스

한국어

HyperAI초신경

Multi Hypotheses 3D Human Pose Estimation On

평가 지표

Average MPJPE (mm)

Average PMPJPE (mm)

Using 2D ground-truth joints

평가 결과

이 벤치마크에서 각 모델의 성능 결과

모델 이름	Average MPJPE (mm)	Average PMPJPE (mm)	Using 2D ground-truth joints	Paper Title	Repository
GFPose (HPJ2D-000, S=200)	35.6	30.5	16.9	GFPose: Learning 3D Human Pose Prior with Gradient Fields
Li et al.	73.9	44.3	-	Weakly Supervised Generative Network for Multiple 3D Human Pose Hypotheses
cGNF xlarge w Lsample	48.5	-	-	Multi-hypothesis 3D human pose estimation metrics favor miscalibrated distributions
MDN	52.7	42.6	-	Generating Multiple Hypotheses for 3D Human Pose Estimation with Mixture Density Network
D3DP	35.4	-	No	Diffusion-Based 3D Human Pose Estimation with Multi-Hypothesis Aggregation
Sharma et al.	46.8	37.3	-	Monocular 3D Human Pose Estimation by Generation and Ordinal Ranking
GraphMDN	46.2	36.3	-	GraphMDN: Leveraging graph structure and deep learning to solve inverse problems	-
MHEntropy	-	36.8	-	MHEntropy: Entropy Meets Multiple Hypotheses for Pose and Shape Recovery	-
GFPose (HPJ2D-010, S=200)	35.1	-	-	GFPose: Learning 3D Human Pose Prior with Gradient Fields
cGNF w Lsample	53	-	-	Multi-hypothesis 3D human pose estimation metrics favor miscalibrated distributions

0 of 10 row(s) selected.