ORDNet: Capturing Omni-Range Dependencies for Scene Parsing

TL;DR

ORDNet is a novel neural network architecture that captures short, middle, and long-range dependencies in scene images, significantly improving scene parsing accuracy across multiple benchmarks.

Contribution

The paper introduces a Middle-Range branch and a Reweighed Long-Range branch to fill the gap between local and global dependencies in scene parsing models.

Findings

Outperforms state-of-the-art on PASCAL Context

Achieves superior results on COCO Stuff

Sets new benchmarks on ADE20K

Abstract

Learning to capture dependencies between spatial positions is essential to many visual tasks, especially the dense labeling problems like scene parsing. Existing methods can effectively capture long-range dependencies with self-attention mechanism while short ones by local convolution. However, there is still much gap between long-range and short-range dependencies, which largely reduces the models' flexibility in application to diverse spatial scales and relationships in complicated natural scene images. To fill such a gap, we develop a Middle-Range (MR) branch to capture middle-range dependencies by restricting self-attention into local patches. Also, we observe that the spatial regions which have large correlations with others can be emphasized to exploit long-range dependencies more accurately, and thus propose a Reweighed Long-Range (RLR) branch. Based on the proposed MR and RLR branches, we build an Omni-Range Dependencies Network (ORDNet) which can effectively capture short-, middle- and long-range dependencies. Our ORDNet is able to extract more comprehensive context information and well adapt to complex spatial variance in scene images. Extensive experiments show that our proposed ORDNet outperforms previous state-of-the-art methods on three scene parsing benchmarks including PASCAL Context, COCO Stuff and ADE20K, demonstrating the superiority of capturing omni-range dependencies in deep models for scene parsing task.