Spherical CNNs (2018)

Summary

[Submitted on 30 Jan 2018 (v1), last revised 25 Feb 2018 (this version, v3)] Title:Spherical CNNs View a PDF of the paper titled Spherical CNNs, by Taco S. Cohen and 3 other authors View PDF Abstract:Convolutional Neural Networks (CNNs) have become the method of choice for learning problems involving 2D planar images. However, a number of problems of recent interest have created a demand for models that can analyze spherical images. Examples include omnidirectional vision for drones, robots, and autonomous cars, molecular regression problems, and global weather and climate modelling. A naive application of convolutional networks to a planar projection of the spherical signal is destined to fail, because the space-varying distortions introduced by such a projection will make translational weight sharing ineffective. In this paper we introduce the building blocks for constructing spherical CNNs. We propose a definition for the spherical cross-correlation that is both expressive and rotation-equivariant. The spherical correlation satisfies a generalized Fourier theorem, which allows us to compute it efficiently using a generalized (non-commutative) Fast Fourier Transform (FFT) algorithm. We demonstrate the computational efficiency, numerical accuracy, and effectiveness of spherical CNNs applied to 3D model recognition and atomization energy regression. Submission history From: Taco Cohen [view email] [v1] Tue, 30 Jan 2018 18:28:30 UTC (1,942 KB) [v2] Thu, 8 Feb 2018 08:06:34 UTC (1,942 KB) [v3] Sun, 25 Feb 2018 13:43:49 UTC (1,942 KB)