Skip to content

Ochuman

Name

OCHuman Dataset

Description

13360 annotated human instances (19 keypoints) within 5081 images.

Important: Please keep in mind to manually download AND unzip the dataset from here. In order for the IceData parser to work on this data, you will need the ochuman.json annotations file and the images directory containing the actual images. You will get both unzipping the folder obtained here.

Check out the parser in action at ochuman.ipynb.