Bag-of-visual-words and spatial extensions for land-use classification

Abstract

We investigate bag-of-visual-words (BOVW) approaches to land-use classification in high-resolution overhead imagery. We consider a standard non-spatial representation in which the frequencies but not the locations of quantized image features are used to discriminate between classes analogous to how words are used for text document classification without regard to their order of occurrence. We also consider two spatial extensions, the established spatial pyramid match kernel which considers the absolute spatial arrangement of the image features, as well as a novel method which we term the spatial co-occurrence kernel that considers the relative arrangement. These extensions are motivated by the importance of spatial structure in geographic data.The methods are evaluated using a large ground truth image dataset of 21 land-use classes. In addition to comparisons with standard approaches, we perform extensive evaluation of different configurations such as the size of the visual dictionaries used to derive the BOVW representations and the scale at which the spatial relationships are considered.We show that even though BOVW approaches do not necessarily perform better than the best standard approaches overall, they represent a robust alternative that is more effective for certain land-use classes. We also show that extending the BOVW approach with our proposed spatial cooccurrence kernel consistently improves performance.

Very Deep Convolutional Neural Networks for Complex Land Cover Mapping Using Multispectral Remote Sensing Imagery

“…Previous studies reported the superiority of fine-tuning relative to full-training for classification of very high resolution aerial imagery, although full-training was found to be more accurate relative to fine-tuning for classification of multi-spectral satellite data [26,45]. In particular, Nogueira et al (2017) evaluated the efficiency of fine-tuning and full-training strategies of some well-known deep CNNs (e.g., AlexNet and GoogLeNet) for classification of three well-known datasets, including UCMerced land-use [46], RS19 dataset [47], and Brazilian Coffee Scenes [48]. The fine-tuning strategy yielded a higher accuracy for the first two datasets, likely due to their similarity with the ImageNet dataset, which was originally used for training deep CNNs.…”

Multiview Deep Learning for Land-Use Classification

“…The UC Merced land-use dataset [1] is investigated, which is a set of aerial ortho-imagery with a 0.3048 m pixel resolution extracted from United States Geological Survey (USGS) national maps. The UC Merced dataset has been used as a benchmark for land-use classifier evaluation in numerous publications [1]- [7].…”

“…The UC Merced dataset has been used as a benchmark for land-use classifier evaluation in numerous publications [1]- [7].…”

“…The objective of this research was to design a deep convolutional neural network (DCNN) for the UC Merced land-use dataset [1], a dataset compiled in 2010 and used as a benchmark in several land-use classification studies [1]- [7]. The challenge is to optimize classification accuracy by finding a proper selection of DCNN hyper-parameters, which are defined as the DCNN settings, such as the architecture design, convolutional filter bank specifications, pooling layer specifications, and learning rate and momentum values, excluding the learned neuron weights and biases.…”

Using surfaces and surface relations in an Early Cognitive Vision system

“…After defining rather simple, and usually task specific feature representations, some kind of classifiers are learned as, e.g., in bag of words approaches (see, e.g., [11,63,87]) or in many industrial applications (see, e.g., [86]). Such systems can lead to rather impressive results for specific tasks, but-as discussed aboveface the inherent limitations of flat architectures.…”