Semantic Image Segmentation with Task-Specific Edge Detection Using CNNs and a Discriminatively Trained Domain Transform

TL;DR

This paper introduces a fast, learnable edge-preserving filtering method using domain transform in CNN-based semantic segmentation, replacing dense CRFs to improve boundary accuracy efficiently.

Contribution

It proposes a novel, end-to-end trainable domain transform approach that learns task-specific edges for improved semantic segmentation boundary accuracy.

Findings

Domain transform filtering is several times faster than dense CRF inference.

The method achieves comparable segmentation accuracy with improved boundary delineation.

Learning task-specific edges enhances segmentation performance.

Abstract

Deep convolutional neural networks (CNNs) are the backbone of state-of-art semantic image segmentation systems. Recent work has shown that complementing CNNs with fully-connected conditional random fields (CRFs) can significantly enhance their object localization accuracy, yet dense CRF inference is computationally expensive. We propose replacing the fully-connected CRF with domain transform (DT), a modern edge-preserving filtering method in which the amount of smoothing is controlled by a reference edge map. Domain transform filtering is several times faster than dense CRF inference and we show that it yields comparable semantic segmentation results, accurately capturing object boundaries. Importantly, our formulation allows learning the reference edge map from intermediate CNN features instead of using the image gradient magnitude as in standard DT filtering. This produces task-specific edges in an end-to-end trainable system optimizing the target semantic segmentation quality.