ZEISS Knowledge Base
Help
ZEISS Knowledge Base
Search

Intellesis Segmentation Models

Creating and Training an Intellesis Segmentation Model

For a good training result always note the following:

  1. The more accurately you perform the labeling the better the result will be. You can start with only a few annotations and then check the result for problematic areas where you should add more labels to refine the labeling.
  2. Fewer, more precise labels result in more accurate models than numerous imprecise labels.
  3. Take care to also label some areas which contain edges of objects and transitions between two classes.
  4. Adopt an iterative approach: review segmentation and training outcomes before labeling large quantities of pixels.
  5. Try to label roughly the same amount of pixels per class.
  6. Do not label very large homogenous areas.
  1. You have completed the general preparations.
  2. You have licensed the AI Toolkit and activated it under Tools > Toolkit Manager.
  1. On the Analysis tab, in the Intellesis Segmentation tool, click and select New.
  2. A text field is displayed.
  3. Enter a name and a description for the new model and click .
  4. A new empty model is created.
  5. Click Start Training.
  6. The user interface for training opens.
  7. In the Right Tool Area under Open Images click Import Images.
  8. Select the image for training from the file system and click Open.
  9. If the image contains more than one channel, the Select Channel dialog opens.
  10. For Training Mode, select either Single Channel or Multispectral. In case of single channel, select the channel you want to use for training. Otherwise all channels of the image will be used for training the model.
  11. Training mode and channel are selected.
  12. Click OK.
  13. All images you import need to have the same channel structure and pixel type of the initially imported image. When importing multiple images, the channel selection applies to all images.
  14. The image is displayed in the list. Note that all imported images will be included in your training model.
  15. Select the image from the list.
  16. The image is displayed in the Center Screen Area. Note that at a later stage you can add more images via Import Images to refine the training.
  17. In the Left Tool Area, define the classes based on your image and segmentation goals. You can specify the number of classes needed for your task. When initiating a new model, two predefined classes (Object and Background) are available by default. If you click on Add Class a new class is added. You can rename these classes by a double-click and entering a new name. Note that you must not use the name Root for one of your classes as this a reserved keyword from the image analysis.
  18. Move the courser inside the image and start labeling the areas which you want to assign to the selected class. To label within the image simply hold down the left mouse key and move the mouse.
  19. After labeling a few areas with different classes, click Train & Segment.
  20. The software starts the training. The system tries to automatically recognize other areas of the same class. Depending on the image, the pixel classification can take a while. When finished the image has the additional channel Seg (menation) containing the segmentation preview.
  21. If you are not satisfied with the result, you can label more details of the corresponding classes. For this you can zoom into the image or change the brush size of the courser. The more accurately you label the different classes within the image, the better the recognition will be. When you finish the labeling, you have to click on Train & Segment again. You can repeat that process until you are satisfied with the segmentation result.
    Note that at this point as a result you will only see a pseudo segmented image and only the area visible in the main window is segmented (max. area 5000x5000 px). The full segmentation of an image/data set is performed on the Processing tab by using the trained model with the Intellesis Segmentation function.
Trained model with refined labeling and pseudo segmentation results
Trained model with refined labeling and pseudo segmentation results

Importing an Intellesis Segmentation Model

  1. You have completed the general preparations.
  2. You have a trained model or a pre-trained neural network available for import.
  3. You are on the Analysis tab.
  1. In the Intellesis Segmentation tool, click and select Import.
  2. A file browser opens.
  3. In the file browser, select the model file or your network from the file system. The network can also be imported by selecting the respective JSON file.
  4. Click Open.
  5. The model is now available in the dropdown list.
  6. Select the model and click Start Training to work with the model, e.g. if you want to train more details.
    If you want to use the model for image processing switch to the Processing tab. In the Intellesis Segmentation processing function you can select the imported model and apply it to the desired images/data sets.

Exporting an Intellesis Segmentation Model

  1. You have created and selected a model for Intellesis segmentation.
  1. On the Analysis tab, in the Intellesis Segmentation tool, click and select Export.
    If you want to export the full model containing all images select Export with Images, see Exporting an Intellesis Segmentation Model with Images.
  2. Select the file location and click Save.
  1. The trained segmentation model is exported. Note that only the model file is exported, intended for segmentation tasks or image analysis setups, not for further training.

Exporting an Intellesis Segmentation Model with Images

  1. You have selected an Intellesis segmentation model with images.
  1. On the Analysis tab, in the Intellesis Segmentation tool, click and select Export With Images.
  2. The file explorer opens.
  3. Navigate to the folder where you want to store the training model and click Save.
  1. The trained segmentation model, along with all images used for training, is exported.

Renaming an Intellesis Segmentation Model

  1. You have selected a trained Intellesis segmentation model.
  1. On the Analysis tab, in the Intellesis Segmentation tool, click and select Rename.
  2. The Model field is editable.
  3. Enter a new name for the model and save it.

Creating an Image Analysis Setting from an Intellesis Segmentation Model

  1. On the Analysis tab, in the Intellesis Segmentation tool, click and select Create Analysis Setting.
  2. The file browser opens with the default location for image analysis settings.
  3. If you want to change the location, select a new path for saving the setting. Click Save.
  1. You have now created and saved a setting which you can use for 2D image analysis, see Performing an Image Analysis Using an Intellesis Segmentation Model.

ON THIS PAGE

PRODUCTS
ZENX Ray SoftwareEM SoftwareSolutions Lab
ABOUT ZEISS
AboutCareerNewsroomCompliance
Social Media
LinkedInFacebookYouTubeXFlickrInstagram
Impressum
Carl-Zeiss-Strasse 22
73447 Oberkochen
Germany
Legal