Advances in scene classification of remotely sensed high resolution images and the existing datasets

Abstract

Research on Scene classification of remotely sensed images has shown a significant improvement in the recent years as it is used in various applications such as urban planning, urban mapping, management of natural resources, precision agriculture, detecting targets etc. The recent advancement of intelligent earth observation system has led to the generation of high resolution remote sensing images in terms of spatial, spectral and temporal resolutions which in turn helped the researchers to improve the performance of Land Use Land Class (LULC) Classification Techniques to a higher level. With the usage of different deep learning architecture and the availability of various high resolution image datasets, the field of Remote Sensing Scene Classification of high resolution (RSSCHR) images has shown tremendous improvement in the past decade. In this paper we present the different publicly available datasets, various scene classification methods and the future research scope of remotely sensed high resolution images.