IPSDK Explorer Artificial Intelligence modules for Segmentation

IPSDK Explorer Artificial Intelligence modules for Segmentation

In addition to traditional image analysis and processing solutions, IPSDK also provides artificial intelligence (AI) solutions.

In fact, it offers two innovative and revolutionary modules based on Machine Learning techniques: Smart Segmentation and Smart Classification. Although technologically highly advanced, these new tools have been designed to be easily used by all our users, even novices.

Moreover, it offers the possibility of generating learning models in the IPSDK Explorer graphical interface, using easy-to-use interactive tools. These models can then be used in the interface or in scripts to rapidly process large quantities of images.

Smart Segmentation : Pixel classification

Smart Segmentation is used to automatically classify pixels or voxels in an image.

Firstly, after a brief, user-friendly manual segmentation learning stage, the Machine Learning tool swiftly defines the rules for classifying image pixels.

Next, this training enables the Machine Learning algorithm to generate a model that can classify the entire image automatically. This model can also be applied to segment other similar images.

Introduction to the SMART Segmentation module available in the IPSDK software.

This video demonstrates how to build a model using the interactive learning tool to classify pixels in an image automatically and simply.

The algorithm, based on Random Forest techniques, identifies the discriminant measures needed to classify objects according to the user-provided training.

Super pixel/Voxel Segmentation : Classification of blocks of pixels automatically pre-cut into homogeneous zones

As an alternative to Smart Segmentation, IPSDK Explorer offers Super Pixel Segmentation for processing thicker objects. This module automatically classifies pixels or voxels in an image based on a pre-cut into homogeneous zones.

First of all, it begins by grouping zones of similar pixels. Then, a few of these super-pixels are associated with each class. This training enables the Machine Learning algorithm to generate a model to automatically classify the entire image. This model can also segment other similar images.

This process allows objects to be outlined quickly and efficiently.

Presentation of the Super-Pixel segmentation module in IPSDK software, version 3.2.

This video demonstrates how to generate a regular image split by aligning the super-pixel borders with object edges, grouping pixels into homogeneous zones.

Next, it shows how to build a model using the interactive learning tool to classify super-pixels automatically and easily.

The algorithm, based on Random Forest techniques, identifies the discriminating measures needed for classification, following the user’s training.