Researchers have developed two new models, Mask2Former and OneFormer, for universal image segmentation. This task involves dividing an image into its constituent parts or objects. The models are based on the Transformer architecture and use attention mechanisms to focus on specific regions of the image. According to the developers, Mask2Former achieves state-of-the-art results on several benchmark datasets, while OneFormer is more efficient but still competitive. --- Why it matters: These models matter because they can improve the accuracy and efficiency of image segmentation tasks in various applications, such as autonomous vehicles, medical imaging, and robotics. Source: https://huggingface.co/blog/mask2former