Efficient piecewise training of deep structured models for semantic segmentation

TL;DR

This paper introduces a deep structured model for semantic segmentation that leverages contextual information through patch-patch and patch-background interactions, employing efficient training methods to achieve state-of-the-art results.

Contribution

It proposes a novel deep structured model with CRF-based pairwise potentials and efficient piecewise training for improved semantic segmentation performance.

Findings

Achieved 78.0 IOU on PASCAL VOC 2012

Set new state-of-the-art on multiple datasets

Demonstrated effectiveness of patch-based contextual modeling

Abstract

Recent advances in semantic image segmentation have mostly been achieved by training deep convolutional neural networks (CNNs). We show how to improve semantic segmentation through the use of contextual information; specifically, we explore `patch-patch' context between image regions, and `patch-background' context. For learning from the patch-patch context, we formulate Conditional Random Fields (CRFs) with CNN-based pairwise potential functions to capture semantic correlations between neighboring patches. Efficient piecewise training of the proposed deep structured model is then applied to avoid repeated expensive CRF inference for back propagation. For capturing the patch-background context, we show that a network design with traditional multi-scale image input and sliding pyramid pooling is effective for improving performance. Our experimental results set new state-of-the-art performance on a number of popular semantic segmentation datasets, including NYUDv2, PASCAL VOC 2012, PASCAL-Context, and SIFT-flow. In particular, we achieve an intersection-over-union score of 78.0 on the challenging PASCAL VOC 2012 dataset.