Login Register

Auto-login on future visits

Forgot your password?

Easily build powerful classifiers.

Choose the best classifier for the task. Easily train accurate and generalizing classifiers. Consistent syntax and parameter naming let you focus on your problem, not on the tools.

State-of-the-art classifiers such as neural networks, random forest and support vector machines in perClass
visualizing classifier decisions in feature space

See what you trained.

Visualization helps our understanding. To see the partitioning of your feature space may give you an important insight into performance of your classifier. And yes, with perClass you can view it also in multi-D spaces. Learn more

Detect anything.

Building a detector is not a rocket science. With one command, you just provide a target class and a model to use. That's it. Quickly build one-class classifiers or powerful detectors you can tune later with ROC. Start thinking about detectors as building blocks. Because that's what they really are in your system.

Optimize detector using ROC analysis
Cascaded classifier with detector followed by classifier.

Cascade classifiers. In one line.

We rarely solve industrial problems with a single classifier. You don't only find a defect but also classify which one it is. You don't tackle a complex problem with one model. Instead, you split it into several simpler ones and use different features and classifier in each. Cascading helps you to connect all the pieces into one classification system. Because it is so simple you don't need to think how to use it. You just do.

Performance tuning made simple.

Just training a classifier is not enough. You must make sure you don't loose the rare defect or discard too much of the expensive mineral. That's why we need to optimize classifier performance.

perClass comes with easy to use tools that allow you to understand the capability of your classifier and tune it to perform exactly as you need.

Tune classifier performance using interactive ROC plot.