Designing deep CNN models based on sparse coding for aerial imagery: a deep-features reduction approach

Abstract

Traditional methods focus on low-level handcrafted features representations and it is difficult to design a comprehensive classification algorithm for remote sensing scene classification problems. Recently, convolutional neural networks (CNNs) have obtained remarkable performance outcomes, setting several remote sensing benchmarks. Furthermore, direct applications of UAV remote sensing images that use deep convolutional networks are extremely challenging given high input data dimensionality with relatively small amounts of available labelled data. We, therefore, propose a CNN approach to scene classification that architecturally incorporates sparse coding (SC) technique for dimension reduction to minimize overfitting. Outcomes were compared with principal component analysis (PCA) and global average pooling (GAP) alternatives that use fully connected layer(s) in pre-trained CNN architecture(s) to minimize overfitting. SC was used to encode deep features extracted from the last convolutional layer of pre-trained CNN models by using different features maps in which deep features had been converted into low-dimensional SC features. These same sparse-coded features were concatenated by means of different pooling techniques to obtain global image features for scene classification. The proposed algorithm outperformed current state-of-the-art algorithms based on handcrafted features. When using our own UAVbased dataset and existing datasets, it was also exceptionally efficient computationally when learning data representations, producing a 93.64% accuracy rate..

Crop identification by massive processing of multiannual satellite imagery for EU common agriculture policy subsidy control

“…In the experiment on Section 3, we have used the well-known dense neural networks (DNNs), which are suitable with the proposal and yielded good enough results. Other alternatives, for example, Random Forest (Chunhui et al, 2017) or more complex CNNs such as VGG (Abdul et al, 2019), used in similar projects may also be appropriate, but the good results of the DNNs have helped us to test the method and carry out the experiment.…”

UAV-Based Slope Failure Detection Using Deep-Learning Convolutional Neural Networks

“…Although these classification algorithms yield acceptable results for remote sensing classification problems, CNNs have recently been used to obtain state-of-the-art results in computer vision and some limited remote sensing applications (i.e., object detection and scene classification). However, using UAV remotely sensed images for the CNNs still presents a challenge because of the limited availability of labelled data [23]. The multiple layers in feed-forward CNNs are able to provide critical feature representations of an image in a hierarchical manner, which allows them to distinguish the visual laws in the feed-forwarded image from any expert-designed complex ruleset [29].…”

“…Using a dropout of 0.5 is useful for fully connected layers to increase the generalization of the approach and avoid overfitting, which result in better transferability of the approach. Dropout is considered as a regularization method which use during training and it randomly dropped the connections among the network for reducing overfitting [23]. The kernel sizes and the number of feature maps were selected on the basis of our input window sizes.…”

“…Over the last decade deep-learning models, and in particular CNNs, have been successfully used for classification and segmentation of remotely sensed imagery [23][24][25], for scene annotation, and for a broad range of object detection applications [26]. Although there are some ways of using deep learning approaches in an unsupervised mode, CNNs are supervised machine learner in which masses of labelled sample patches are used to feed learnable filters, in order to minimize the applied loss function [27].…”

Rapid mapping of landslides in the Western Ghats (India) triggered by 2018 extreme monsoon rainfall using a deep learning approach

“…The recent advancements in the performance of computing platforms have resulted in the development of several machine learning models, including deep learning methods (DLMs). Of the developed DLMs, the deep convolutional neural networks (DCNNs) have been especially widely used for classification and segmentation of satellite images and object detection (Du et al 2019;Qayyum et al 2019).…”