Open Images is a dataset of ~9 million URLs to images that have been annotated with labels spanning over 6000 categories. This page aims to provide the download instructions and mirror sites for Open Images Dataset. Please visit the project page for more details on the dataset .
Prerequisite: Gmail or Gmail associated account
CVDF hosts image files that have bounding boxes annotations in the Open Images Dataset V4. The images are split into train (1,743,042), validation (41,620), and test (125,436) sets. The images are rescaled to 1024x768 resolution with total size 561GB. The images can be directly downloaded into a local directory or a Google Cloud storage bucket. Please sign up with your Gmail or Gmail asscociated account here to request access. After you submit the request form, we will grant READ access to your mail account for the storage bucket:
gs://open-images-dataset
You can either download the images to a storage bucket or a local directory with the following procedures:
- install gsutil
- gcloud auth login [your_mail_account]
- gsutil -m rsync -r gs://open-images-dataset/train [target_dir/train] (513GB)
gsutil -m rsync -r gs://open-images-dataset/validation [target_dir/validation] (12GB)
gsutil -m rsync -r gs://open-images-dataset/test [target_dir/test] (36GB)
The target_dir can be a local directory or a Google Cloud storage bucket.
Prerequisite: Google Cloud Platform account
In this section, we describe the procedures to download all images in the Open Images Dataset to a Google Cloud storage bucket. We recommend to use the user interface provided in the Google Cloud storage console for the task.
Google Storage provides a "storage transfer" function to transfer online files into a storage bucket. This function can be used to transfer images from original urls into user's storage bucket. CVDF prepares the tsv files that contain all image urls in Open Images Dataset for the transfer. The step-by-step instructions are described in Creating and Managing Transfers with the Console. The size of the whole dataset is around 18TB. Please note that user needs to pay for hosting the dataset on Google Cloud storage after downloading it. The hosting price can be found on Google Cloud Storage Pricing.
The tsv file for the train set (partitioned into 10 files):
https://storage.googleapis.com/cvdf-datasets/oid/open-images-dataset-train[0-9].tsv
The tsv file for the validation set:
https://storage.googleapis.com/cvdf-datasets/oid/open-images-dataset-validation.tsv
The tsv file for the test set:
https://storage.googleapis.com/cvdf-datasets/oid/open-images-dataset-test.tsv
React
Typescript
Django
Laravel
Facebook
Microsoft
Google
Alibaba
D3
Tencent