Comparing the Deep Convolutional Neural Network Models for Identification of Aerial Image Land Use

Aerial vehicles, such as drones, have emerged as indispensable tools for capturing high-resolution images, commonly called aerial images. These captivating visuals hold immense potential in land usage identification, revolutionizing remote sensing, geospatial analysis, and urban planning. Leveraging aerial images has proven to be a game-changer, offering an efficient and cost-effective approach to monitoring and scrutinizing land usage and cover changes over time. The wealth of information from these images has paved the way for valuable insights across diverse fields, ranging from agriculture and forestry to environmental management. The paper proposes four Pretrained DCNNs, which are VGG16, DenseNet121, ResNet50, and Mobile Net. In this method, we would be fusing the auto-extracted feature of Pretrained DCNN, which is used for performance accuracy of land usage identification. In this, we have considered eight machine learning algorithms with three machine learning ensembles considered.