OptimEyes Developer version 25.6.5.0 is now available. This version includes an updated ML Tool with the following enhancements:

  • Added support for ML models that can be used to generate images, such as image segmentation models. For example, the ML Tool can now be used to easily generate a binary mask or grayscale image for each class into which a segmentation model can segment an image.
  • Added the ability to apply the softmax function, normalization, and multiplication to the results provided by ML models. These features are useful in cases such as when ML models return raw logits instead of probabilities, or return values that are not in the desired range, such as probabilities between 0 and 1 when the desired output is a grayscale image with values from 0 to 255.
  • The ML Tool now saves and restores more of the input values connected with each ML model, making it easier to configure models that you have used before for new tasks.
  • Added default values for more settings, enabling you to get an ML model up and running without having to attend to as many details.
  • Added the ability to list only operations whose names do not contain any slashes. These will almost always be the operations whose names you will need to specify for inputs and outputs.
  • Other performance optimizations and minor bug fixes.