Unsupervised Image Semantic Segmentation through Superpixels and Graph Neural Networks

TL;DR

This paper introduces an unsupervised image segmentation method combining mutual information maximization, superpixels, and graph neural networks to learn meaningful image representations without labeled data.

Contribution

It proposes a novel end-to-end approach integrating GNNs with superpixels and MIM, which models long-range pixel interactions and improves segmentation accuracy.

Findings

Outperforms state-of-the-art methods on four datasets

Provides qualitative and quantitative improvements

Models interactions between distant pixels effectively

Abstract

Unsupervised image segmentation is an important task in many real-world scenarios where labelled data is of scarce availability. In this paper we propose a novel approach that harnesses recent advances in unsupervised learning using a combination of Mutual Information Maximization (MIM), Neural Superpixel Segmentation and Graph Neural Networks (GNNs) in an end-to-end manner, an approach that has not been explored yet. We take advantage of the compact representation of superpixels and combine it with GNNs in order to learn strong and semantically meaningful representations of images. Specifically, we show that our GNN based approach allows to model interactions between distant pixels in the image and serves as a strong prior to existing CNNs for an improved accuracy. Our experiments reveal both the qualitative and quantitative advantages of our approach compared to current state-of-the-art methods over four popular datasets.