• Simple User Interface for Labelling and Training
• Integration into ZEN Measurement Framework
• Support for multi-dimensional datasets
• Machine-Learning Tool for Pixel Classification powered by Python
• Scikit-Learn
• Tensorflow
• Dask
• Client/Server Architecture with REST-API
• Engineered Default Feature Sets (CPU)
• 25 or 33 Features
• Neural Network (vgg19) Layers for Feature Extraction (GPU)
• 64, 128 (red. 50) or 256 (red. 70) Features for 1st, 2nd or 3rd layer
• Random Forest Classifier for Pixel Classification (CPU)
• Option to download pre-trained DNNs (Deep neural Networks) for specific sample types (subject of change)
• Post Processing: Conditional Random Field (CRF)
• IP-Functions for creating masks and confidence maps
• Integration into the OAD scripting environment (Developer Toolkit) for advanced automation