Optimized convolutional neural network for land cover classification via improved lion algorithm

Dependable land cover data are required to aid in the resolution of a broad spectrum of environmental issues. Land cover classification at a broad scale has been carried out using data from traditional ground-based information from the Advanced Very High-Resolution Radiometer. From the merits as well as demerits of the existing works discussed in the literature, this article seeks to establish a novel technique for automatic, fast, as well as precise land cover classification deploying remote sensing data. The proposed approach follows feature extraction and classification stages. From input information, the statistical characteristics are extracted as well as they are subjected to classification via optimized deep convolutional neural network. Particularly, the convolutional layers are optimized for by means of a new Proposed Lion Algorithm with a new Cub pool Update (PLACU) approach. The established model is the advanced level of the standard lion algorithm. The superiority of the established technique is determined by the extant techniques regarding positive and negative metrics. The accuracy of the work that is being presented (PLACU) is superior to the existing methods like Dragonfly algorithm, Jaya algorithm, sea lion optimization, and lion algorithm techniques by 20%, 15%, 112%, and 10%, respectively.