MPII Human Pose#

Format specification#

The original MPII Human Pose Dataset is available here.

Supported annotation types:

Bbox
Points

Supported attributes:

center (a list with two coordinates of the center point of the object)
scale (float)

Import MPII Human Pose Dataset#

A Datumaro project with an MPII Human Pose Dataset source can be created in the following way:

datum project create
datum project import --format mpii <path/to/dataset>

It is also possible to import the dataset using Python API:

import datumaro as dm

mpii_dataset = dm.Dataset.import_from('<path/to/dataset>', 'mpii')

MPII Human Pose Dataset directory should have the following structure:

dataset/
├── mpii_human_pose_v1_u12_1.mat
├── 000000001.jpg
├── 000000002.jpg
├── 000000003.jpg
└── ...

Export to other formats#

Datumaro can convert an MPII Human Pose Dataset into any other format Datumaro supports. To get the expected result, convert the dataset to a format that supports bounding boxes or points.

There are several ways to convert an MPII Human Pose Dataset to other dataset formats using CLI:

datum project create
datum project import -f mpii <path/to/dataset>
datum project export -f voc -o ./save_dir -- --save-media

or

datum convert -if mpii -i <path/to/dataset> \
    -f voc -o <output/dir> -- --save-media

Or, using Python API:

import datumaro as dm

dataset = dm.Dataset.import_from('<path/to/dataset>', 'mpii')
dataset.export('save_dir', 'voc')

Examples#

Examples of using this format from the code can be found in the format tests