Label Refinement Network for Coarse-to-Fine Semantic Segmentation

TL;DR

This paper introduces a label refinement network that improves semantic image segmentation by progressively refining labels from coarse to fine resolutions using multi-stage supervision.

Contribution

It presents a novel coarse-to-fine network architecture with multi-stage loss functions for enhanced semantic segmentation performance.

Findings

Effective pixel-wise dense labeling on standard datasets

Improved segmentation accuracy over baseline models

Multi-resolution supervision enhances detail refinement

Abstract

We consider the problem of semantic image segmentation using deep convolutional neural networks. We propose a novel network architecture called the label refinement network that predicts segmentation labels in a coarse-to-fine fashion at several resolutions. The segmentation labels at a coarse resolution are used together with convolutional features to obtain finer resolution segmentation labels. We define loss functions at several stages in the network to provide supervisions at different stages. Our experimental results on several standard datasets demonstrate that the proposed model provides an effective way of producing pixel-wise dense image labeling.