Label classification (Weka)

Labels in label images can be classified according to their shape and size or to intensity measurements derived from a corresponding intensity image.

How to do label classification

Open your data set. Start the CLIJx-Assistant and follow these steps:

• Your dataset
  • CLIJx-Assistant Starting point
    • Thresholding (e.g. Otsu)
      • Connected Components Labeling
        • Weka Label Classifier

In the dialog of the classifier, make sure to select the right intensity input image.

Download video

Furthermore, choose the right features, depending on your desired analysis (shape, intensity), by clicking on the “Features…” button:

The model will then be saved to the specified file location when you hit the “Train” button.

Note: For technical reasons, the model files are not compatible with traditional Trainable Weka Segmentation.

Back to CLIJx-Assistant

Imprint