RT info:eu-repo/semantics/article
T1 Real-time semantic segmentation for fisheye urban driving images based on ERFNet
A1 Bergasa Pascual, Luis Miguel
A1 Saez Contreras, Álvaro
A1 López Guillén, María Elena
A1 Romera Carmena, Eduardo
A1 Tradacete Ágreda, Miguel
A1 Gómez Huélamo, Carlos
A1 Egido Sierra, Javier del
K1 Fisheye
K1 Intelligent vehicles
K1 CNN (Convolutional Neural Network)
K1 Deep Learning
K1 Distortion
K1 Electrónica
K1 Electronics
AB The interest in fisheye cameras has recently risen in the autonomous vehicles field, as they areable to reduce the complexity of perception systems while improving the management of dangerous driving situations. However, the strong distortion inherent to these cameras makes the usage of conventional computer vision algorithms difficult and has prevented the development of these devices. This paper presents a methodology that provides real-time semantic segmentation on fisheye cameras leveraging only synthetic images. Furthermore, we propose some Convolutional Neural Networks (CNN) architectures based on Efficient Residual Factorized Network (ERFNet) that demonstrate notable skills handling distortion and a new training strategy that improves the segmentation on the image borders. Our proposals are compared to similar state-of-the-art works showing an outstanding performance and tested in an unknown real world scenario using a fisheye camera integrated in an open-source autonomous electric car, showing a high domain adaptation capability.
PB MDPI
SN 1424-8220
YR 2019
FD 2019-01-25
LK http://hdl.handle.net/10017/43126
UL http://hdl.handle.net/10017/43126
LA eng
NO Ministerio de Economía y Competitividad
DS MINDS@UW
RD 19-abr-2024