Evaluation of Off-The-Shelf CNNs for the Representation of Natural Scenes with Large Seasonal Variations - Archive ouverte HAL Access content directly
Reports (Research Report) Year : 2017

Evaluation of Off-The-Shelf CNNs for the Representation of Natural Scenes with Large Seasonal Variations

Amandine Gout
  • Function : Author
Yann Lifchitz
  • Function : Author
Jérémy Fix
Cédric Pradalier

Abstract

This paper focuses on the evaluation of deep convolutional neural networks for the analysis of images of natural scenes subjected to large seasonal variation as well as significant changes of lighting conditions. The context is the development of tools for long-term natural environment monitoring with an autonomous mobile robot. We report various experiments conducted on a large dataset consisting of a weekly survey of the shore of a small lake over two years using an autonomous surface vessel. This dataset is used first in a place recognition task framed as a classification problem, then in a pose regression task and finally the internal features learned by the network are evaluated for their representation power. All our results are based on the Caffe library and default network structures where possible.
Fichier principal
Vignette du fichier
bmvc_dl_seasonal.pdf (3.11 Mo) Télécharger le fichier
Origin : Files produced by the author(s)
Loading...

Dates and versions

hal-01448091 , version 1 (27-01-2017)
hal-01448091 , version 2 (09-02-2017)

Licence

Attribution

Identifiers

  • HAL Id : hal-01448091 , version 2

Cite

Amandine Gout, Yann Lifchitz, Titouan Cottencin, Quentin Groshens, Jérémy Fix, et al.. Evaluation of Off-The-Shelf CNNs for the Representation of Natural Scenes with Large Seasonal Variations. [Research Report] UMI 2958 GeorgiaTech-CNRS; CentraleSupélec UMI GT-CNRS 2958 Université Paris-Saclay. 2017. ⟨hal-01448091v2⟩
497 View
213 Download

Share

Gmail Facebook X LinkedIn More