Normalized Cut Loss for Weakly-supervised CNN Segmentation

TL;DR

This paper introduces a normalized cut loss function for weakly-supervised CNN segmentation that improves training quality by evaluating seed labels and pixel consistency, reducing the gap with fully supervised methods.

Contribution

It proposes a novel loss combining seed-specific cross-entropy with a normalized cut criterion evaluated efficiently via bilateral filtering.

Findings

Enhanced segmentation quality with weak supervision

Normalized cut loss improves training stability

Approach achieves results close to fully supervised methods

Abstract

Most recent semantic segmentation methods train deep convolutional neural networks with fully annotated masks requiring pixel-accuracy for good quality training. Common weakly-supervised approaches generate full masks from partial input (e.g. scribbles or seeds) using standard interactive segmentation methods as preprocessing. But, errors in such masks result in poorer training since standard loss functions (e.g. cross-entropy) do not distinguish seeds from potentially mislabeled other pixels. Inspired by the general ideas in semi-supervised learning, we address these problems via a new principled loss function evaluating network output with criteria standard in "shallow" segmentation, e.g. normalized cut. Unlike prior work, the cross entropy part of our loss evaluates only seeds where labels are known while normalized cut softly evaluates consistency of all pixels. We focus on normalized cut loss where dense Gaussian kernel is efficiently implemented in linear time by fast Bilateral filtering. Our normalized cut loss approach to segmentation brings the quality of weakly-supervised training significantly closer to fully supervised methods.