| Abstract |
Computational intelligence (CI) is a group of nature-simulated computational models and processes for addressing difficult real-life problems. The CI is useful in the UAV domain as it produces efficient, precise, and rapid solutions. Besides, unmanned aerial vehicles (UA V) developed a hot research topic in the smart city environment. Despite the benefits of UAVs, security remains a major challenging issue. In addition, deep learning (DL) enabled image classification is useful for several applications such as land cover classification, smart buildings, etc. In this view, this study proposes novel meta-heuristics with a deep learning-driven secure UAV image classification (MDLS-UAVIC) model in a smart city environment. The major purpose of the MDLS-UAVIC algorithm is to securely encrypt the images and classify them into distinct class labels. The proposed MDLS-UAVIC model follows a two-stage process: encryption and image classification. The encryption technique for image encryption effectively encrypts the UAV images. Next, the image classification process involves an Xception-based deep convolutional neural network for the feature extraction process. Finally, shuffled shepherd optimization (SSO) with a recurrent neural network (RNN) model is applied for UAV image classification, showing the novelty of the work. The experimental validation of the MDLS-UA VIC approach is tested utilizing a benchmark dataset, and the outcomes are examined in various measures. © 2023 IEEE. |
