Generate Datasets

Autogenerated datasets consist of randomly generated noise in images for testing purposes and do not include annotations. Below is an example of a noise image:


NOTE: Generated datasets are only in the ImageNet format and suitable only for generic or classification type models.

NOTE: Because autogenerated datasets are made of noise images and do not contain annotations, they cannot be used for accuracy measurement and calibration.

On the Create Configuration page, find the table with datasets:


To generate a new dataset, click the Autogenerate Dataset in ImageNet format button. Fill out the Autogenerate Dataset form that appears on a new page:

Parameter Values Explanation
Number of Images to Generate [1; 2000]A number of .npy images to generate for the dataset
Image Dimensions [1; 1920] Image height and width in pixels

After you have entered all the required parameters, click the Generate button to generate a dataset. You are automatically directed back to the datasets table, where the Status column shows the progress bar and status of the generation.


To cancel the generation process, press a red stop sign in the Action column.

Once the generation completes, the dataset indicates a green tick sign.

To remove an imported dataset from the list, click the bin icon in the Action column.

Select dataset in the list and proceed to select an environment.

See Also