COVID-19 pandemic is the main reason people must wear face masks in public places. Traditionally, officers monitor the use of face masks in the public area manually. However, monitoring masks using manual techniques is challenging in a crowded spot. Thus, we propose a face mask detection based on Generative Adversarial Networks (GAN) through the learning model to accelerate mask detection accurately and quickly. To construct our detection model, we collect the dataset, conduct pre-processing, and train the model by tuning multiple parameters to obtain the highest accuracy and tiny loss. The experimental results can produce D_Loss = 0.0032 and G_Loss = 7.3296. Therefore, the proposed model can be a promising solution for mask detection issues.