Land Use Classification in Remote Sensing Images by Convolutional Neural Networks

By Marco Castelluccio, Giovanni Poggi, Carlo Sansone, Luisa Verdoliva. 2015

632 citations recorded by the source corpus.

We explore the use of convolutional neural networks for the semantic classification of remote sensing scenes. Two recently proposed architectures, CaffeNet and GoogLeNet, are adopted, with three different learning modalities. Besides conventional training from scratch, we resort to pre-trained networks that are only fine-tuned on the target data, so as to avoid overfitting problems and reduce design time. Experiments on two remote sensing datasets, with markedly different characteristics, testify on the effectiveness and wide applicability of the proposed solution, which guarantees a significant performance improvement over all state-of-the-art references.

Subfields: Atmospheric Science, Computer Vision and Pattern Recognition, Ecology, Media Technology

How researchers see this paper

The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.

In Atmospheric Science

In Computer Vision and Pattern Recognition

In Ecology

In Media Technology

Citations in the corpus

EuroSAT: A Novel Dataset and Deep Learning Benchmark for Land Use and Land Cover Classification (2017, methodology): “The presented work extensively evaluated all proposed machine learning methods and concluded that deep CNNs outperformed non-deep learning approaches on the considered datasets [6], [15], [17], [27].”
Large patch convolutional neural networks for the scene classification of high spatial resolution imagery (2016, background): “8,42 However, no published works have discussed the practical design of general CNN architectures that can make the best use of the currently available small HSR-RS scene datasets.”
A Spatial-Temporal Attention-Based Method and a New Dataset for Remote Sensing Image Change Detection (2020, methodology): “In light of the good performance in computer vision, deep-CNN-based methods have been widely applied in remote sensing tasks [25], such as land-use classification [39], semantic segmentation [40,41], image super-resolution [42], object detection [43,44] and change detection [5].”
Remote Sensing Image Scene Classification Using CNN-CapsNet (2019, methodology): “As for CNN, two representative CNN models (VGG‐16 and Inception‐V3) fully trained on the ImageNet dataset are used as initial feature map extractors, considering their popularity in the remote sensing field [25,27,28,35,53,56,59].”
Target-Adaptive CNN-Based Pansharpening (2017, background): “have been applied to remote sensing problems [37]–[39], and several papers have been proposed to address pansharpen-”
Introducing Eurosat: A Novel Dataset and Deep Learning Benchmark for Land Use and Land Cover Classification (2018, methodology): “Accordingly, we use the state-of-the-art deep CNN models GoogleNet [6] and ResNet50 [7] for the classification of the introduced dataset and the 2http://madm.dfki.de/downloads popular remote sensing datasets UCM [2], AID [8], SAT-6 [1] and BCS [4].”
Deep Feature Fusion for VHR Remote Sensing Scene Classification (2017, methodology): “The other is to use pretrained CNNs and then fine-tune them on VHR data set [22], [23].”
Simultaneous extraction of roads and buildings in remote sensing imagery with convolutional neural networks (2017, background): “…2014; Makantasis et al., 2015; Paisitkriangkrai et al., 2015; Kampffmeyer Michael and Jenssen, 2015; Jiang et al., 2015; Vakalopoulou et al., 2015; Castelluccio et al., 2015; Sherrah, 2016; Nogueira et al., 2016), but also in some networks that have been proposed in computer vision, e.g., fully…”