Abstract
Supervised learning techniques are at the center of many tasks in remote sensing. Unfortunately, these methods, especially recent deep learning methods, often require large amounts of labeled data for training. Even though satellites acquire large amounts of data, labeling the data is often tedious, expensive and requires expert knowledge. Hence, improved methods that require fewer labeled samples are needed. We present MSMatch, the first semi-supervised learning approach competitive with supervised methods on scene classification on the EuroSAT benchmark dataset. We test both RGB and multispectral images and perform various ablation studies to identify the critical parts of the model. The trained neural network achieves state-of-the-art results on EuroSAT with an accuracy that is between 1.98% and 19.76% better than previous methods depending on the number of labeled training examples. With just five labeled examples per class we reach 94.53% and 95.86% accuracy on the EuroSAT RGB and multispectral datasets, respectively. With 50 labels per class we reach 97.62% and 98.23% accuracy. Our results show that MSMatch is capable of greatly reducing the requirements for labeled data. It translates well to multispectral data and should enable various applications that are currently infeasible due to a lack of labeled data. We provide the source code of MSMatch online to enable easy reproduction and quick adoption.
Highlights
T HE last decade has seen a momentous increase in the availability of remote sensing data, enhancing the need for efficient image processing and analysis methods using deep learning [1]
We propose MSMatch, a novel approach that builds on recent advances [32] together with recent neural network architectures to tackle the problem of land scene classification, i.e., correctly identifying land use or land cover of satellite or airborne images
The drop in accuracy for all the models is in the range 53%–55%, except for the model trained with 100 labeled images per class, which displays a drop of 61%
Summary
T HE last decade has seen a momentous increase in the availability of remote sensing data, enhancing the need for efficient image processing and analysis methods using deep learning [1] The former is driven by continuously decreasing launch costs, especially for so-called Smallsats (< 500 kg). Since the numbers have been increasing with over 300 launches in 2018 and 2017 [3] Many of these are imaging satellites serving either commercial purposes [4] or related to earth observation programs, such as the European Space Agency’s Copernicus program [2], [5]. This has led to an increase in the availability of large datasets. Such data play a critical role in many tasks ranging from crop modeling [6] to urban analysis [7], [8]
Sign-in/Register to view highlights & summary 
Published Version
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have
Schedule a call