Boundary Guided Context Aggregation for Semantic Segmentation

TL;DR

This paper introduces BCANet, a boundary-guided network that enhances semantic segmentation by using boundary information for context aggregation, leading to improved intra-class consistency and competitive results.

Contribution

The paper proposes a novel boundary-guided context aggregation approach with a multi-scale boundary extractor and an improved non-local module for better semantic understanding.

Findings

Achieves competitive results on Cityscapes and ADE20K datasets.

Enhances intra-class consistency by aggregating context along boundaries.

Demonstrates the effectiveness of boundary-guided context aggregation.

Abstract

The recent studies on semantic segmentation are starting to notice the significance of the boundary information, where most approaches see boundaries as the supplement of semantic details. However, simply combing boundaries and the mainstream features cannot ensure a holistic improvement of semantics modeling. In contrast to the previous studies, we exploit boundary as a significant guidance for context aggregation to promote the overall semantic understanding of an image. To this end, we propose a Boundary guided Context Aggregation Network (BCANet), where a Multi-Scale Boundary extractor (MSB) borrowing the backbone features at multiple scales is specifically designed for accurate boundary detection. Based on which, a Boundary guided Context Aggregation module (BCA) improved from Non-local network is further proposed to capture long-range dependencies between the pixels in the boundary regions and the ones inside the objects. By aggregating the context information along the boundaries, the inner pixels of the same category achieve mutual gains and therefore the intra-class consistency is enhanced. We conduct extensive experiments on the Cityscapes and ADE20K databases, and comparable results are achieved with the state-of-the-art methods, clearly demonstrating the effectiveness of the proposed one.