Skip to main content

Creating datasets

The dataset is a training ready image collection with ground truth data. The LayerNext DatasetManager tool manages and tracks dataset changes over time. A dataset can be created from image collections that are sitting in the DataLake with ground truth Data. Inbuilt augmentation functions in the Datasetmanager can be used to generate additional augmented images if required. Any changes to the dataset such as adding new images can be tracked with version control. With the LayerNext Python SDK, the dataset can be downloaded to a training computer for model training purposes.