The transformation of wireless radio telecommunications in recent times cannot be undeniable with advancement of technologies such as cognitive radio. The Cognitive radio network has been relied upon to solve the problem of spectrum scarcity from the inefficient utilization and dynamic spectrum access as a result of ever-increasing demand to resource allocation of the radio spectrum. The utilization of artificial intelligence and its subsets such as deep learning for spectrum sensing, has been very crucial to the research in cognitive radio and spectrum works. This document offers an extensive summary of Deep Learning (DL) methods employed in Spectrum Sensing, categorized into: Multilayer Perceptrons (MLPs), Convolutional Neural Networks (CNNs), Long Short-Term Memory (LSTM) networks, integrated CNN-LSTM architectures, and Autoencoders (AEs). The studies are examined in terms of the input given to the DL model, the technique used for data collection, the structure of each algorithm, the metrics used for evaluation, the outcomes achieved, and a comparative analysis based on overall characteristics.