Universal Segmentation

Universal segmentation is a challenging computer vision task that aims to segment images into semantic regions, regardless of the task or the domain. It requires the model to learn a wide range of visual concepts and to be able to generalize to new tasks and domains.