Login Register

Auto-login on future visits

Forgot your password?

Newsletter July 2014

Quick image labeling by clustering

Labeling hyperspectral image by cluster analysis in interactive image view.


Labeling image regions is needed when building classifiers. Manual labeling is time consuming and yields small data sets. Cluster analysis offers much faster way of distinguishing image regions.
In this video, you can see a walk-through example labeling a hyper-spectral image by clustering, renaming and merging clusters, training a classifier and applying it to a new image.
Watch this short video to learn more.

In-house perClass training courses

in-house pattern recognition course


Looking for a machine learning course, but find it difficult to leave your office for a week? We bring the training to your office!
We have designed three or five day course to be given in-house. Your entire team gets trained at the same time, in your familiar environment, and, most importantly, focused on your application domain!

perClass course in machine learning November 2014

Mark your calendar!

Upcoming events

The coming autumn is rich with perClass events:

Multi-scale image feature extraction: Practical use-cases


When classifying images, it is useful to extract features at different scales. While small scales describe object details, larger ones capture bigger structures. How to combine them into one data set, so that we can optimally make use of them?
In this tutorial we demonstrate how to compute features at different scales, gather all data in one data set, and, most importantly, pass the labels between the multi-scale data and the original image.

Read more in this knowledge base article.