Decoupling Spherical Reasoning from Dense Prediction for 360 Depth Estimation

2026-09-29

AI Summary

This study addresses the geometric distortion and feature modeling challenges induced by equirectangular projection (ERP) in omnidirectional depth estimation by proposing a novel framework that decouples spherical reasoning from dense prediction. The core innovation lies in performing quasi-uniform node reasoning directly within the native spherical space to eliminate stretching bias—a first in this domain. Specifically, a Fibonacci sphere graph (FSG) is employed to construct a sparse graph structure, which, combined with a spherical context conditioning (SCC) module, enables cross-space feature fusion to guide pixel-level depth prediction. This approach significantly reduces computational overhead while preserving geometric consistency. Extensive experiments on three benchmark datasets demonstrate that the proposed method consistently outperforms existing approaches, validating its superior accuracy and robustness for 360-degree depth estimation tasks.

Abstract

The equirectangular projection (ERP) is widely used for panoramic depth estimation, but its spatially varying distortion makes geometry-consistent feature modeling challenging. We revisit panoramic depth estimation by decoupling contextual modeling in native spherical space from dense ERP prediction. To this end, we propose a Fibonacci Spherical Graph (FSG) as an intermediate reasoning space to lift ERP features onto quasi-uniform Fibonacci nodes on the sphere and capture local and long-range dependencies through complementary spherical neighborhoods. The resulting spherical discretization distributes graph nodes approximately uniformly over the spherical surface, reducing the over-representation of highly stretched regions during relational modeling. Operating on a compact set of Fibonacci nodes also avoids the computational burden of constructing and processing a graph at full ERP resolution. To bridge spherical reasoning and dense prediction, we propose a Spherical Context Conditioning (SCC) module that adaptively modulates dense ERP features with the enhanced spherical representation, allowing spherical context to guide pixel-aligned depth prediction. Extensive experiments on three benchmarks demonstrate that the proposed method consistently achieves superior depth accuracy over existing approaches.

Problem

• 360 depth estimation
• equirectangular projection
• spherical distortion
• panoramic depth estimation
• geometry-consistent feature modeling

Innovation

• 360 Depth Estimation
• Fibonacci Spherical Graph
• Spherical Context Conditioning
• Equirectangular Projection
• Decoupled Reasoning