Double Similarity Distillation for Semantic Image Segmentation

TL;DR

This paper introduces a double similarity distillation framework that enhances the accuracy of compact semantic segmentation networks by capturing pixel and category-level similarities, achieving state-of-the-art results with minimal computational overhead.

Contribution

The paper proposes a novel double similarity distillation framework with pixel-wise and category-wise modules to improve compact segmentation networks without extra parameters.

Findings

Outperforms current state-of-the-art methods on four datasets.

Reduces computational cost while improving accuracy.

Effective across multiple challenging datasets.

Abstract

The balance between high accuracy and high speed has always been a challenging task in semantic image segmentation. Compact segmentation networks are more widely used in the case of limited resources, while their performances are constrained. In this paper, motivated by the residual learning and global aggregation, we propose a simple yet general and effective knowledge distillation framework called double similarity distillation (DSD) to improve the classification accuracy of all existing compact networks by capturing the similarity knowledge in pixel and category dimensions, respectively. Specifically, we propose a pixel-wise similarity distillation (PSD) module that utilizes residual attention maps to capture more detailed spatial dependencies across multiple layers. Compared with exiting methods, the PSD module greatly reduces the amount of calculation and is easy to expand. Furthermore, considering the differences in characteristics between semantic segmentation task and other computer vision tasks, we propose a category-wise similarity distillation (CSD) module, which can help the compact segmentation network strengthen the global category correlation by constructing the correlation matrix. Combining these two modules, DSD framework has no extra parameters and only a minimal increase in FLOPs. Extensive experiments on four challenging datasets, including Cityscapes, CamVid, ADE20K, and Pascal VOC 2012, show that DSD outperforms current state-of-the-art methods, proving its effectiveness and generality. The code and models will be publicly available.