Panoptic segmentation brings together two separate tasks: instance and semantic segmentation. Although they are related, unifying them faces an apparent paradox: how to learn simultaneously instance-specific and category-specific (i.e. instance-agnostic) representations jointly. Hence, state-of-the-art panoptic segmentation methods use complex models with a distinct stream for each task. In contrast, we propose Hierarchical Lov\ász Embeddings, per pixel feature vectors that simultaneously encode instance- and category-level discriminative information. We use a hierarchical Lov\ász hinge loss to learn a low-dimensional embedding space structured into a unified semantic and instance hierarchy without requiring separate network branches or object proposals. Besides modeling instances precisely in a proposal-free manner, our Hierarchical Lov\ász Embeddings generalize to categories by using a simple Nearest-Class-Mean classifier, including for non-instance "stuff" classes where instance segmentation methods are not applicable. Our simple model achieves state-of-the-art results compared to existing proposal-free panoptic segmentation methods on Cityscapes, COCO, and Mapillary Vistas. Furthermore, our model demonstrates temporal stability between video frames.
View on arXiv