NeuralAnnot: Neural Annotator for 3D Human Mesh Training Sets

Introduction

This repo provides 3D pseudo-GTs (SMPL/MANO/FLAME/SMPL-X parameters) of various datasets, obtained by NeuralAnnot (CVPRW 2022 Oral.). We additionally provide 3D pseudo-GTs of SMPL parameters of MSCOCO, obtained by Three Recipes for Better 3D Pseudo-GTs of 3D Human Mesh Estimation in the Wild (CVPRW 2023). You need to install smplx.

Human3.6M

• [data]
• [SMPL parameters]
• [SMPL-X parameters]

MPI-INF-3DHP

• [data]
• [SMPL parameters]
• [SMPL-X parameters]

MSCOCO

• [data]
• [SMPL parameters]
• [MANO parameters]
• [FLAME parameters]
• [SMPL-X parameters (whole body)]
• [SMPL parameters from Three Recipes]
• [SMPL parameters from Two Recipes (without using 3DPW training set)]

MPII 2D Pose Dataset

• [data]
• [SMPL parameters]
• [SMPL-X parameters]

3DPW

• [data]
• [SMPL-X parameters]

CrowdPose

• [SMPL parameters]

FFHQ

• [data]
• [FLAME parameters]

InstaVariety

• [data]
• [SMPL parameters]

InterHand2.6M

• [MANO parameters]

Reference

@InProceedings{Moon_2022_CVPRW_NeuralAnnot,  
author = {Moon, Gyeongsik and Choi, Hongsuk and Lee, Kyoung Mu},  
title = {NeuralAnnot: Neural Annotator for 3D Human Mesh Training Sets},  
booktitle = {Computer Vision and Pattern Recognition Workshop (CVPRW)},  
year = {2022}  
}  

@InProceedings{Moon_2023_CVPRW_3Dpseudpgts,  
author = {Moon, Gyeongsik and Choi, Hongsuk and Chun, Sanghyuk and Lee, Jiyoung and Yun, Sangdoo},  
title = {Three Recipes for Better 3D Pseudo-GTs of 3D Human Mesh Estimation in the Wild},  
booktitle = {Computer Vision and Pattern Recognition Workshop (CVPRW)},  
year = {2023}  
}