K-Net: Towards Unified Image Segmentation

TL;DR

K-Net introduces a unified, learnable kernel-based framework for semantic, instance, and panoptic segmentation, achieving state-of-the-art results with end-to-end training and NMS-free inference.

Contribution

It proposes a novel unified segmentation framework using learnable kernels and a kernel update strategy, simplifying and improving upon specialized existing methods.

Findings

Surpasses previous state-of-the-art on MS COCO panoptic segmentation

Achieves high accuracy on ADE20K semantic segmentation

Offers faster inference speeds comparable to Cascade Mask R-CNN

Abstract

Semantic, instance, and panoptic segmentations have been addressed using different and specialized frameworks despite their underlying connections. This paper presents a unified, simple, and effective framework for these essentially similar tasks. The framework, named K-Net, segments both instances and semantic categories consistently by a group of learnable kernels, where each kernel is responsible for generating a mask for either a potential instance or a stuff class. To remedy the difficulties of distinguishing various instances, we propose a kernel update strategy that enables each kernel dynamic and conditional on its meaningful group in the input image. K-Net can be trained in an end-to-end manner with bipartite matching, and its training and inference are naturally NMS-free and box-free. Without bells and whistles, K-Net surpasses all previous published state-of-the-art single-model results of panoptic segmentation on MS COCO test-dev split and semantic segmentation on ADE20K val split with 55.2% PQ and 54.3% mIoU, respectively. Its instance segmentation performance is also on par with Cascade Mask R-CNN on MS COCO with 60%-90% faster inference speeds. Code and models will be released at https://github.com/ZwwWayne/K-Net/.