Notebook

Datasets¶

This section describes how to install common datasets used for training new models and for computing evaluation scores for entire datasets. In general, these datasets are large and require a computer with a good GPU to train and evaluate in reasonable times.

These datasets are not required to do pose predictions on your own images.

You are unlikely to need all the datasets, so donwload only what you need.

Download COCO data¶

COCO is a great datasets containing many types of annotations, including bounding boxes, 2D poses, etc.

mkdir data-mscoco
cd data-mscoco

wget http://images.cocodataset.org/annotations/annotations_trainval2017.zip
wget http://images.cocodataset.org/annotations/image_info_test2017.zip
unzip annotations_trainval2017.zip
unzip image_info_test2017.zip

mkdir images
cd images
wget http://images.cocodataset.org/zips/val2017.zip
wget http://images.cocodataset.org/zips/train2017.zip
wget http://images.cocodataset.org/zips/test2017.zip
unzip val2017.zip
unzip train2017.zip
unzip test2017.zip

COCO Person Skeletons¶

COCO / kinematic tree / dense:

{image}
:height: "250"

{image}
:height: "250"

{image}
:height: "250"

Created with python3 -m openpifpaf.datasets.constants.

(coco-person-keypoints)=

COCO Person Keypoints¶

In [ ]:

import openpifpaf
openpifpaf.plugin.register()

for i, name in enumerate(openpifpaf.plugins.coco.constants.COCO_KEYPOINTS):
    print(i, name)

In [ ]:

print('associations')
kp_names = openpifpaf.plugins.coco.constants.COCO_KEYPOINTS
for i, (joint1, joint2) in enumerate(openpifpaf.plugins.coco.constants.COCO_PERSON_SKELETON):
    print('{:2d}: {:15s} --> {}'.format(i, kp_names[joint1 - 1], kp_names[joint2 - 1]))

Download MPII data¶

This MPII data is currently not used anywhere.

mkdir data-mpii
cd data-mpii
wget https://datasets.d2.mpi-inf.mpg.de/andriluka14cvpr/mpii_human_pose_v1.tar.gz
wget https://datasets.d2.mpi-inf.mpg.de/andriluka14cvpr/mpii_human_pose_v1_u12_2.zip

Download NightOwls data¶

mkdir data-nightowls
cd data-nightowls
wget http://www.robots.ox.ac.uk/\~vgg/data/nightowls/python/nightowls_validation.json
wget http://www.robots.ox.ac.uk/\~vgg/data/nightowls/python/nightowls_validation.zip
unzip nightowls_validation.zip

In [ ]: