Removing a photo background is easy when the subject is clearly separated from its surroundings. More complicated images, however, can require a different approach. U²-Net is the 176 MB general-purpose background-removal model available in ImageLabs 2.3, giving you an alternative to the lightweight U²-Netp and Silueta models. It is designed for salient-object segmentation, helping identify the main subject so you can create cut-outs and apply creative background effects. Here’s how U²-Net fits into an offline Windows photo-editing workflow.

What Is U²-Net?

U²-Net is a deep-learning model developed for salient-object detection. In simple terms, it analyses an image to identify the object or region that stands out as the main subject.

ImageLabs uses U²-Net for automatic background removal. The model generates a subject mask that distinguishes the foreground from the surrounding background. That mask can then be used to isolate the subject or create other effects without manually tracing its outline.