Geometry Fidelity for Spherical Images

Anders Christensen*, Nooshin Mojab, Khushman Patel, Karan Ahuja, Zeynep Akata, Ole Winther, Mar Gonzalez-Franco, Andrea Colaco

*Corresponding author af dette arbejde
1 Citationer (Scopus)

Abstract

Spherical or omni-directional images offer an immersive visual format appealing to a wide range of computer vision applications. However, geometric properties of spherical images pose a major challenge for models and metrics designed for ordinary 2D images. Here, we show that direct application of Fréchet Inception Distance (FID) is insufficient for quantifying geometric fidelity in spherical images. We introduce two quantitative metrics accounting for geometric constraints, namely Omnidirectional FID (OmniFID) and Discontinuity Score (DS). OmniFID is an extension of FID tailored to additionally capture field-of-view requirements of the spherical format by leveraging cubemap projections. DS is a kernel-based seam alignment score of continuity across borders of 2D representations of spherical images. In experiments, OmniFID and DS quantify geometry fidelity issues that are undetected by FID.

OriginalsprogEngelsk
TitelComputer Vision – ECCV 2024 - 18th European Conference, Proceedings
RedaktørerAleš Leonardis, Elisa Ricci, Stefan Roth, Olga Russakovsky, Torsten Sattler, Gül Varol
Antal sider17
ForlagSpringer
Publikationsdato2025
Sider276-292
ISBN (Trykt)9783031729881
DOI
StatusUdgivet - 2025
Begivenhed18th European Conference on Computer Vision, ECCV 2024 - Milan, Italien
Varighed: 29 sep. 20244 okt. 2024

Konference

Konference18th European Conference on Computer Vision, ECCV 2024
Land/OmrådeItalien
ByMilan
Periode29/09/202404/10/2024
NavnLecture Notes in Computer Science
Vol/bind15138
ISSN0302-9743

Fingeraftryk

Dyk ned i forskningsemnerne om 'Geometry Fidelity for Spherical Images'. Sammen danner de et unikt fingeraftryk.

Citationsformater