Human Pose Forecasting On Human36M

Metrics

Average MPJPE (mm) @ 1000 ms

Average MPJPE (mm) @ 400ms

Results

Performance results of various models on this benchmark

Model Name	Average MPJPE (mm) @ 1000 ms	Average MPJPE (mm) @ 400ms	Paper Title	Repository
UNIMASK-M	112.1	61.6	A Unified Masked Autoencoder with Patchified Skeletons for Motion Synthesis	-
GraphMixer	108.6	56.7	Graph-Guided MLP-Mixer for Skeleton-Based Human Motion Prediction	-
BoM	-	-	Accurate and Diverse Sampling of Sequences based on a "Best of Many" Sample Objective
STARS	-	-	Diverse Human Motion Prediction Guided by Multi-Level Spatial-Temporal Anchors	-
GSPS	-	-	Generating Smooth Pose Sequences for Diverse Human Motion Prediction
ResChunk	81.95	56.41	Multiscale Residual Learning of Graph Convolutional Sequence Chunks for Human Motion Prediction
MotionMixer	111.0	59.3	MotionMixer: MLP-based 3D Human Body Pose Forecasting
TCD	-	-	A generic diffusion-based approach for 3D human pose prediction in the wild
EqMotion	106.9	55.0	EqMotion: Equivariant Multi-agent Motion Prediction with Invariant Interaction Reasoning
MSR-GCN	114.2	62.9	MSR-GCN: Multi-Scale Residual Graph Convolution Networks for Human Motion Prediction	-
HP-GAN	-	-	HP-GAN: Probabilistic 3D human motion prediction via GAN
GCNext	64.7	30.5	GCNext: Towards the Unity of Graph Convolutions for Human Motion Prediction
Sun et al.	109.2	55.5	Towards Accurate Human Motion Prediction via Iterative Refinement	-
STS-GCN	117.0	65.8	Space-Time-Separable Graph Convolutional Network for Pose Forecasting
TP-RNN	-	-	Action-Agnostic Human Pose Forecasting
BC+WGAIL-div	-	-	Imitation Learning for Human Pose Prediction	-
AuxFormer	107	54.1	Auxiliary Tasks Benefit 3D Skeleton-based Human Motion Prediction
Pose-Knows	-	-	The Pose Knows: Video Forecasting by Generating Pose Futures
Forecast LSTM	149.2	80.8	Intention-based Long-Term Human Motion Anticipation
MT-VAE	-	-	MT-VAE: Learning Motion Transformations to Generate Multimodal Human Dynamics

0 of 33 row(s) selected.