fiftyone.zoo.datasets.base.OpenImagesV7Dataset

Open Images V7 is a dataset of ~9 million images, roughly 2 million of which are annotated and available via this zoo dataset.

The dataset contains annotations for classification, detection, segmentation, point labels, and visual relationship tasks for the 600 boxable object classes.

This dataset supports partial downloads:

• You can specify subsets of data to download via the``label_types``, classes, attrs, and max_samples parameters
• You can specify specific images to load via the image_ids parameter

See :ref:`this page <dataset-zoo-open-images-v6>` for more information about partial downloads of this dataset.

Full split stats:

• Train split: 1,743,042 images (513 GB)
• Test split: 125,436 images (36 GB)
• Validation split: 41,620 images (12 GB)

Notes:

• Not all images contain all types of labels
• All images have been rescaled so that their largest dimension is at most 1024 pixels

Example usage:

#
# Load 50 random samples from the validation split
#
# By default, all label types are loaded, including "points"
#

dataset = foz.load_zoo_dataset(
    "open-images-v7",
    split="validation",
    max_samples=50,
    shuffle=True,
)

session = fo.launch_app(dataset)

#
# Load detections, classifications, and points for 25 samples from the
# validation split that contain fedoras and pianos
#
# Images that contain all `label_types` and `classes` will be
# prioritized first, followed by images that contain at least one of
# the required `classes`. If there are not enough images matching
# `classes` in the split to meet `max_samples`, only the available
# images will be loaded.
#
# Images will only be downloaded if necessary
#

dataset = foz.load_zoo_dataset(
    "open-images-v7",
    split="validation",
    label_types=["detections", "classifications", "points"],
    classes=["Fedora", "Piano"],
    max_samples=25,
)

session.dataset = dataset

#
# Download the entire validation split and load detections
#
# Subsequent partial loads of the validation split will never require
# downloading any images
#

dataset = foz.load_zoo_dataset(
    "open-images-v7",
    split="validation",
    label_types=["detections"],
)

session.dataset = dataset

Dataset size

561 GB

Source

https://storage.googleapis.com/openimages/web/index.html