Training Models with the Segmentation Wizard

With the Segmentation Wizard, you simply paint the different features of interest within a limited subset of your image data and then train models to identify objects according to a predefined set of rules. The most promising model can then be fine-tuned and published for repeated segmentation tasks. This empowers any user to achieve robust and reproducible segmentation results that are not influenced by user bias. The main tasks for training semantic segmentation models with the Segmentation Wizard are labeling classes in frames, choosing a model generation strategy, and then training the models within the selected strategy.