AI-based methods

Utilizes machine learning algorithms to classify regions based on a trained Intellesis (Link) model (based on random forest pixel classification) or an imported deep learning network, e.g., ONNX models or models trained on arivis Cloud.

Use case: Labeling image regions defined by classes in a previously trained model.

For information on labeling/annotating images to train a model on arivis Cloud, see Annotation of Images for Model Training on arivis Cloud.

For general information about arivis Cloud, see arivis Cloud Overview.

Post-processing tools in the 2D Toolkit in ZEN and ZEN core allow further refinement of semantic segmentation results, such as fill holes / binary operations and separation algorithms such as watershed.

This method leverages deep learning models for distinguishing and separating individual objects, even when they touch or overlap.

Use case: Provide precise identification of objects in complex images.

Intellesis Trainable Segmentation, part of the AI Toolkit license, is a powerful machine learning tool integrated into the ZEN and ZEN core platforms, enabling you to perform trainable class segmentation.

Use case: Classify image regions based on specific features, such as intensity and texture by training the models directly within the software.

The Intellesis Training UI provides an intuitive interface for creating and refining models using conventional machine learning techniques, such as random forests. You can then apply these models to classify pixels into predefined classes, facilitating accurate segmentation tailored to the unique characteristics of a dataset.