Squeeze-and-Attention Networks for Semantic Segmentation

TL;DR

This paper introduces SANet, a novel architecture with squeeze-and-attention modules that enhance semantic segmentation by focusing on pixel-group and pixel-wise features, achieving state-of-the-art results.

Contribution

The paper proposes a new squeeze-and-attention module that improves segmentation by capturing spatial-channel dependencies and multi-scale context.

Findings

Achieves 83.2% mIoU on PASCAL VOC without COCO pre-training.

Achieves 54.4% mIoU on PASCAL Context, setting a new state-of-the-art.

Effectively models pixel-group and pixel-wise attention for better segmentation.

Abstract

The recent integration of attention mechanisms into segmentation networks improves their representational capabilities through a great emphasis on more informative features. However, these attention mechanisms ignore an implicit sub-task of semantic segmentation and are constrained by the grid structure of convolution kernels. In this paper, we propose a novel squeeze-and-attention network (SANet) architecture that leverages an effective squeeze-and-attention (SA) module to account for two distinctive characteristics of segmentation: i) pixel-group attention, and ii) pixel-wise prediction. Specifically, the proposed SA modules impose pixel-group attention on conventional convolution by introducing an 'attention' convolutional channel, thus taking into account spatial-channel inter-dependencies in an efficient manner. The final segmentation results are produced by merging outputs from four hierarchical stages of a SANet to integrate multi-scale contexts for obtaining an enhanced pixel-wise prediction. Empirical experiments on two challenging public datasets validate the effectiveness of the proposed SANets, which achieves 83.2% mIoU (without COCO pre-training) on PASCAL VOC and a state-of-the-art mIoU of 54.4% on PASCAL Context.