StyleGAN-Human: A Data-Centric Odyssey of Human Generation

SHHQ Dataset

ECCV 2022

Jianglin Fu1*, Shikai Li1*, Yuming Jiang2, Kwan-Yee Lin1, Chen Qian1,
Chen Change Loy2, Wayne Wu1,3 †, Ziwei Liu2
1SenseTime Research, 2S-Lab, Nanyang Technological University, 3Shanghai AI Laboratory
Back to main page


Dataset Overview

SHHQ is a dataset with high-quality full-body human images in a resolution of 1024 × 512.
For now, SHHQ-1.0 with 40K images is released! More data will be released in the later versions.

Preview

Hover mouse over the image for a zoomed-in view.

  Click here to download the preview version with 100 images randomly sampled from SHHQ-1.0.

Annotation

SHHQ-1.0 comes with machine-calculated segmentation masks, manually annotated human parsing labels, and 2D body keypoints.

Download instruction for full SHHQ-1.0 dataset: