ImageNet is a large image database widely used in deep learning and computer vision research. The dataset provides images for thousands of nouns and follows the hierarchical structure defined by WordNet.

This manual explains how to use the ImageNet module. 

• Available Datasets
• Access Requirements
• Access to the Module

Available Datasets

The following ImageNet datasets are available as modules:

• ImageNet Winter 2021 (full dataset)

• ImageNet Large-Scale Visual Recognition Challenge (ILSVRC) 2012

Access Requirements

Access to the ImageNet modules on Claix must be requested.

To obtain access, you must:

1. Have an active ImageNet account.

2. Accept the ImageNet license terms.

Then send a mail to the IT-ServiceDesk with:

• A screenshot of your ImageNet account page

• The screenshot must show your access key and email address

Once your request has been approved, access to the ImageNet modules will be enabled.

Access to the Module

To list all available ImageNet module versions, run:

module spider ImageNet

To load a specific version of the dataset (e.g. Winter21_whole), use:

module load ImageNet/Winter21_whole

After loading the module, the ImageNet dataset can be accessed via the environment variable:

$R_IMAGENET