perClass 5.2 release brings interactive confusion matrix
We enter to 2018 with a new perClass 5.2 release. It brings interactive confusion matrix that significantly simplifies tuning multi-class classifiers and exploring their performance capabilities. There is also support for C-weighting in SVM which allows us to explicitly fix importance of certain classes and significantly faster training of RBF networks scalable to very large problems. Happy new year 2018 to everyone!
Partnership with Cubert brings perClass to real-time spectral cameras
Happy to share that PR Sys Design and Cubert GmbH established a partnership tighly integrating perClass into Cubert ecosystem of real-time spectral cameras.
Spring’17 course fully-booked, registration for June course open
Happy to share that our March’17 course is fully booked (as the two previous courses). We have opened the registration for the Summer edition in the week 26-30 June 2017 in Delft, NL.
Spring 2017 course scheduled in the week 20-24 March 2017
perClass 5.0 release brings Deep Learning
Happy to announce perClass 5 with support for deep convolutional networks! The new sddeepnet command allows you to design architecture and train a deep network in one step. A GUI with training progress is available so that network behaviour can be understood and architecture improved. Other improvements include image feature extraction and upgrade to SVM optimizer.
You can check the tutorial video
More info in Release notes
Autumn perClass course scheduled for 24-28 October’16 in Delft
The Autumn course “perClass for R&D Specialists” is scheduled in the week 24-28 October 2016 in Delft, The Netherlands
Sumer 2016 perClass course will take place in Eindhoven in the week 27 June-1July (register now)
Happy to announce Summer 2016 perClass course that will take place on the High-Tech Campus in Eindhoven in the week 27 June to 1 July. The registration is now open.
4.8 release brings new image feature extractors and zoom in image figures
Happy to announce availability of perClass 4.8 that brings support for zooming in image figures (also with a convenient keyboard +/- shortcuts, that do not need mode change), new image feature extractors (orientation histograms and gray-level morphology) and more! More information in Release notes. Enjoy!
Spring perClass course scheduled for the week 4-8 April 2016
The spring perClass Course “Machine Learning for R&D Specialists” will be in the week 4-8 April 2016. Registrations are open.
4.6 release with uint8 buffer support, improved ROC and filter bank features
4.6 release brings significantly simpler setting of a current op.point on ROC characteristic with a combination of a constrain and performance maximization. New enhancement is a support for direct classifier execution on uint8 or uint16 buffers (think applying classifiers to raw image buffers without content casting - very cool!). There are also several new convolutional filter banks for local image feature extraction and multiple fixes. More information in Release notes. Enjoy!