Videos

Human3.6M

Introduced by Ionescu et al. in Human3.6m: Large scale datasets and predictive methods for 3D human sensing in natural environments

The Human3.6M dataset is one of the largest motion capture datasets, which consists of 3.6 million human poses and corresponding images captured by a high-speed motion capture system. There are 4 high-resolution progressive scan cameras to acquire video data at 50 Hz. The dataset contains activities by 11 professional actors in 17 scenarios: discussion, smoking, taking photo, talking on the phone, etc., as well as provides accurate 3D joint positions and high-resolution videos.

Source: Space-Time Representation of People Based on 3D Skeletal Data: A Review

Homepage

Benchmarks

Add a new result Link an existing benchmark

Task	Dataset Variant	Best Model
3D Human Pose Estimation	Human3.6M	BCP+VHA R152 384x384
Monocular 3D Human Pose Estimation	Human3.6M	MotionBERT
Weakly-supervised 3D Human Pose Estimation	Human3.6M	AdaptPose
Human Pose Forecasting	Human3.6M	EqMotion
Unsupervised 3D Human Pose Estimation	Human3.6M	ElePose
Multi-Hypotheses 3D Human Pose Estimation	Human3.6M	GFPose
Video Prediction	Human3.6M	IAM4VP
Unsupervised Human Pose Estimation	Human3.6M	B-KinD
Human action generation	Human3.6M	Kinetic-GAN
3D Absolute Human Pose Estimation	Human3.6M	GnTCN
3D Pose Estimation	Human3.6M	TesseTrack
Human Part Segmentation	Human3.6M	Hulk
Root Joint Localization	Human3.6M	GnTCN
Pose Retrieval	Human3.6M	Pr-VIPE
2D Pose Estimation	Human3.6M	UniHCP
3D Human Pose Estimation in Limited Data	Human3.6M	PoseDA