Masked-attention Mask Transformer for Universal Image Segmentation
This reduces research effort and improves performance for computer vision researchers and practitioners working on image segmentation.
The paper tackles the problem of designing specialized architectures for different image segmentation tasks by introducing Mask2Former, a universal architecture that outperforms specialized models, achieving state-of-the-art results such as 57.8 PQ on COCO for panoptic segmentation.
Image segmentation is about grouping pixels with different semantics, e.g., category or instance membership, where each choice of semantics defines a task. While only the semantics of each task differ, current research focuses on designing specialized architectures for each task. We present Masked-attention Mask Transformer (Mask2Former), a new architecture capable of addressing any image segmentation task (panoptic, instance or semantic). Its key components include masked attention, which extracts localized features by constraining cross-attention within predicted mask regions. In addition to reducing the research effort by at least three times, it outperforms the best specialized architectures by a significant margin on four popular datasets. Most notably, Mask2Former sets a new state-of-the-art for panoptic segmentation (57.8 PQ on COCO), instance segmentation (50.1 AP on COCO) and semantic segmentation (57.7 mIoU on ADE20K).