Cam-CONVS: Camera-aware multi-scale convolutions for single-view depth

Facil, J.M. (Universidad de Zaragoza) ; Ummenhofer, B. ; Zhou, H. ; Montesano, L. (Universidad de Zaragoza) ; Brox, T. ; Civera, J. (Universidad de Zaragoza)

Resumen: Single-view depth estimation suffers from the problem that a network trained on images from one camera does not generalize to images taken with a different camera model. Thus, changing the camera model requires collecting an entirely new training dataset. In this work, we propose a new type of convolution that can take the camera parameters into account, thus allowing neural networks to learn calibration-aware patterns. Experiments confirm that this improves the generalization capabilities of depth prediction networks considerably, and clearly outperforms the state of the art when the train and test images are acquired with different cameras.

Idioma: Inglés

DOI: 10.1109/CVPR.2019.01210

Año: 2019

Publicado en: Proceedings - IEEE Computer Society Conference on Computer Vision and Pattern Recognition 2019 (2019), 11818-11827
ISSN: 1063-6919

Factor impacto SCIMAGO: 13.396 - Software - Computer Vision and Pattern Recognition

Financiación: info:eu-repo/grantAgreement/ES/DGA-FSE/T45-17R
Financiación: info:eu-repo/grantAgreement/EC/H2020/688007/EU/A gardening robot for rose, hedge and topiary trimming/TrimBot2020
Financiación: info:eu-repo/grantAgreement/ES/MINECO/DPI2015-67275

Tipo y forma: Article (PostPrint)

Área (Departamento): Área Ingen.Sistemas y Automát. (Dpto. Informát.Ingenie.Sistms.)
Área (Departamento): Área Lenguajes y Sistemas Inf. (Dpto. Informát.Ingenie.Sistms.)