Evaluation of Split-Brain Autoencoders for High-Resolution Remote Sensing Scene Classification

Self-supervised methods are interesting for remote sensing because there are not many human labeled datasets available, but there is practically unlimited amount of data that can be used for self-supervised learning. In this paper we analyze the use of split-brain autoencoders in the context of remo...

Full description

Saved in:
Bibliographic Details
Published in2018 International Symposium ELMAR pp. 67 - 70
Main Authors Stojnic, Vladan, Risojevic, Vladimir
Format Conference Proceeding
LanguageEnglish
Published Croatian Society Electronics in Marine - ELMAR 01.09.2018
Subjects
Aerial image classification
Colorization
Feature extraction
Image color analysis
Quantization (signal)
Remote sensing
Self-supervised learning
Support vector machines
Task analysis
Training
Online AccessGet full text

Cover

More Information
Summary:Self-supervised methods are interesting for remote sensing because there are not many human labeled datasets available, but there is practically unlimited amount of data that can be used for self-supervised learning. In this paper we analyze the use of split-brain autoencoders in the context of remote sensing image classification. Weinvestigate the importance of training set size, choice of color space and size of the model to the classification accuracy. We show that even with small amount of unlabeled training images, if we finetune the weights learned by the autoencoder, we can achieve almost state of the art results of 89.27% on AID dataset.
ISBN:9531842442
9789531842440
DOI:10.23919/ELMAR.2018.8534634