Your First dataspan.ai Project – Workflow Overview

Here’s an overview of the stages of creating your first dataspan.ai Project. Detailed steps are provided in the next chapter – Using dataspan.ai.

Stage 1, Datasets – Importing Data into dataspan.ai

What: Import datasets of images containing foreground objects to be inserted into background images into dataspan.ai, including metadata, such as bounding boxes and/or segmentation masks.

Where: Import datasets of background images (canvases) into dataspan.ai. These are the images into which foreground objects will be inserted. Detailed steps are provided in Step 5, Datasets – Importing Data into dataspan.ai.

Stage 2, Concepts – Creating dataspan.ai’s Image Generative Model

Feed dataspan.ai’s generative AI algorithm with the object images to insert by providing suitable images and possibly describing them textually.

dataspan.ai inserts variations of these foreground objects into the background images that you specify here.

You can then review examples of the synthetic data generated by dataspan.ai and select the best Presets (hyperparameters) for the dataspan.ai model to use.

Detailed steps are provided in Step 6, Creating Synthetic Data Concepts.

Stage 3, Batches – Generating Data at Scale

Batches represent the application of what the dataspan.ai model learned during the Concepts stage at scale in order to generate numerous images into the background images of your choice. This means that running a Batch inserts the foreground objects that the dataspan.ai model learned in the Concepts stage into the background images that you specify here.

Detailed steps are provided in Step 7, Generating Batches – Applying Concepts at Scale.