Integrated solution for industrial machine learning

perClass is a software tool that enables R&D specialists to develop classification solutions With a Matlab-based perClass Toolbox, you can understand your problem, interactively work with data, choose best features, train classifiers and optimize their performance with ROC analysis.  With perClass SDK, you can use perClass Runtime to embed classifiers in your applications running without Matlab.


perClass Mira is a user-friendly application for interpretation of spectral images. It allows rapid construction of classification solutions without programming or machine learning expertise.

  • User loads pre-recorded spectral images.
  • Defines classes of interest (e.g. normal/rotten potato).
  • User paints class labels in the respective image areas.
  • Mira automatically select the best machine learning approach and provides solution (decisions on the image).
  • User may inspect decisions over more images and improve the solution, if needed.
  • Solution may be executed on new images in a production mode.

See all benefits

perClass is built for R&D specialists using classification as a tool in their project. To get started, basic Matlab scripting skills are needed. However, you don’t need to have PhD in machine learning!*

To embed classifiers in custom applications, you or your developers need to call a dynamic library (DLL) using your favourite language (C++, C#, Visual Basic, LabView, Matlab compiler or others).

Designed from the ground up as a cross-platform solution, perClass allows you to train your classifiers on a powerful 64-bit Windows workstation and deploy them to a tiny disk-less Linux-based device. Whether on Windows, Linux or Mac, perClass offers the same outstanding performance. The choice is yours.

Our products are thoroughly checked prior to delivery!

We will respond to your service request, promptly!

Deliveries are made in time!