Reviewing Datasets

This page shows each of the Datasets imported into the current Project, showing its creation date and the number of images that each contains. The information shown may differ slightly according to the Type of the Dataset that was imported: Image Folder, Bounding Boxes or Segmentation Masks.

To display the details of a Dataset, double-click on it.

To return to the Projects page, click .

Tip: We recommend reviewing the images imported into the Dataset to verify that they appear as expected in dataspan.ai. You can also display a visualization of the bounding boxes defined in the annotations files. dataspan.ai overlays these bounding boxes on the displayed images.

Image Folder Dataset

Images folders typically do not contain annotation files describing bounding boxes, segmentation masks and classes. Therefore, this type of Dataset is used for background images, meaning the canvas on which the foreground objects will be inserted. For example, as shown below:

If a down arrow appears on the right, you can click it to display additional information about the Dataset.

Image Samples

Clicking on an image displays a zoomed-in view with tools for inspecting it .

Bounding Box Dataset

This type of Dataset is typically suitable for foreground images, meaning the objects to later be inserted into the background images (such as a defect). The location in the image of the object to be inserted is typically indicated by the associated annotation files (bounding boxes and segmentation masks).

To display information about the classes, click the down arrow .

This example shows that of the 30 images in this Dataset, 15 are live_knot class, 15 are resin class, and 0 have No class.

The Show Bounding Boxes switch enables you to visualize the bounding boxes defined in the annotations files so that they appear overlaid on the displayed images. For example, as shown below:

Note: You can rename a Dataset without affecting functionality. You can also remove a Dataset and import a different one in its place, as needed.

Failed to Import

If a Datasets fails to import (for example, if its format is incorrect), it appears in the Datasets page as follows:

You can click on it to investigate the reason. For example, as shown below: