Seg-HGNN: Unsupervised and Light-Weight Image Segmentation with Hyperbolic Graph Neural Networks

TL;DR

Seg-HGNN introduces a lightweight hyperbolic graph neural network for image segmentation, leveraging hyperbolic embeddings to outperform existing unsupervised methods with fewer parameters and faster processing.

Contribution

The paper proposes a novel hyperbolic GNN architecture for image segmentation that is unsupervised, lightweight, and more effective than current methods.

Findings

Outperforms current best unsupervised methods on multiple datasets.

Uses less than 7.5k trainable parameters for efficient performance.

Achieves approximately 2 images per second on standard GPUs.

Abstract

Image analysis in the euclidean space through linear hyperspaces is well studied. However, in the quest for more effective image representations, we turn to hyperbolic manifolds. They provide a compelling alternative to capture complex hierarchical relationships in images with remarkably small dimensionality. To demonstrate hyperbolic embeddings' competence, we introduce a light-weight hyperbolic graph neural network for image segmentation, encompassing patch-level features in a very small embedding size. Our solution, Seg-HGNN, surpasses the current best unsupervised method by 2.5\%, 4\% on VOC-07, VOC-12 for localization, and by 0.8\%, 1.3\% on CUB-200, ECSSD for segmentation, respectively. With less than 7.5k trainable parameters, Seg-HGNN delivers effective and fast ($\approx 2$ images/second) results on very standard GPUs like the GTX1650. This empirical evaluation presents compelling evidence of the efficacy and potential of hyperbolic representations for vision tasks.