The rapid popularity of Unmanned Aerial Vehicles (UAVs) or drones is due to its great promise for various commercial and recreational applications.  Moreover, technological advancements has made UAVs both affordable and versatile, it is able to perform task considered difficult and dangerous with  high mobility, low cost and safely. Despite these advantageous applications and future potentials, UAVs are regrettably also used for malicious activities  such as cyber-attack (i.e., eavesdropped the mobile phone of unsuspecting users), terrorism, drug trafficking and violation of security of restricted or  sensitive areas. In order to overcome these aforementioned challenges, various UAV detection and identification techniques were developed by  researchers in the past. Conventional techniques have inherent weakness such as poor performance in the presence of low visibility, less effective in an  environment with high ambient noise and are also adversely affected by weather conditions. All these and more constitute major setback in UAV detection and identification. Similarly, RF sensing employed are usually based on traditional machine learning and statistical methods which are less  effective in their capacity to successfully capture the rich information present in complex unstructured RF signal. To this end, a deep learning approach is  proposed as a possible solution that is capable of achieving high accuracy in detection with less training time. To verify the effectiveness of our proposed  convolutional neural network architecture, it was evaluated on the popular open source dataset, the DRONE RF dataset. Our empirical results  demonstrates that our proposed CNN model outperformed state-of- the -art results using the same DRONE RF dataset.