The rapid increase in number of vehicles in recent times has adversely affected the travel time, traffic blocks, and accidents. Random search for a parking space contributes around 30% of city traffic which costs a significant amount of time and energy. Hence, smart parking solutions that detect and allocate vacant parking spaces in real-time are essential to minimize this traffic congestion. In this paper, we propose a novel method to detect the occupancy status of an outdoor parking space using Long Term Evolution (LTE)-based Channel State Information (CSI) and Convolutional Neural Network (CNN). This supervised classification method can provide real-time status of the occupancy. In this study, we analyze the performance of the proposed method by comparing with other CSI-based localization techniques. Through numerical results, we show that the proposed method outperforms the state-of-the-art techniques. © 2020 IEEE.