AI Toolkit

Learn more about segmentation.

Segmentation is an image processing method that identifies and separates different regions of interest.

• Semantic segmentation (pixel-based segmentation) is a form of segmentation that assigns the same pixel value to all pixels in an image corresponding to a particular class.
• Instance segmentation (object-based segmentation) is another form of segmentation that assigns a unique pixel value to all pixels in an image corresponding to each instance or object of a class. The values assigned range from 0, which represents the background, to N, which corresponds to the total number of objects present in the image.

Semantic segmentation involves the classification of individual pixels and is the most commonly used task in microscopy applications. If you are unsure about which approach to take, semantic segmentation is a good choice. However, if you need to segment individual objects in addition to pixels, instance segmentation is recommended for special applications. It's important to note that instance segmentation requires more training data and longer training times compared to semantic segmentation.

You can read our blog post, or watch this video to learn more about different ways of achieving segmentation.

Have questions? We're here to help.

Have a question about segmentation? Feel free to reach out to our support team.