Hyperbolic Image Segmentation

2022/03/11 by Mina GhadimiAtigh, GhadimiAtigh, Mina, Julian Schoep +7 · 7 citations
Computer Science · #Topological and Geometric Data Analysis #Digital Image Processing Techniques #Medical Image Segmentation Techniques

paper · pdf · doi:10.48550/arxiv.2203.05898

Abstract

For image segmentation, the current standard is to perform pixel-level optimization and inference in Euclidean output embedding spaces through linear hyperplanes. In this work, we show that hyperbolic manifolds provide a valuable alternative for image segmentation and propose a tractable formulation of hierarchical pixel-level classification in hyperbolic space. Hyperbolic Image Segmentation opens up new possibilities and practical benefits for segmentation, such as uncertainty estimation and boundary information for free, zero-label generalization, and increased performance in low-dimensional output embeddings.

Cited by

• PointCHR: Point Cloud Analysis via Curvature-Aware Hyperbolic Rectification
• H²em: Learning Hierarchical Hyperbolic Embeddings for Compositional Zero-Shot Learning
• LIHE: Linguistic Instance-Split Hyperbolic-Euclidean Framework for Generalized Weakly-Supervised Referring Expression Comprehension
• PHyCLIP: ℓ₁-Product of Hyperbolic Factors Unifies Hierarchy and Compositionality in Vision-Language Representation Learning
• Talk in Pieces, See in Whole: Disentangling and Hierarchical Aggregating Representations for Language-based Object Detection
• Hyperbolic Multimodal Representation Learning for Biological Taxonomies
• Hyperbolic Cycle Alignment for Infrared-Visible Image Fusion

Related

• Seg-HGNN: Unsupervised and Light-Weight Image Segmentation with Hyperbolic Graph Neural Networks
• Hyperbolic Deep Learning in Computer Vision: A Survey
• Hyperbolic Image Embeddings
• Hyperbolic Geometry in Computer Vision: A Survey
• Geometry-aware Distance Measure for Diverse Hierarchical Structures in Hyperbolic Spaces