Face recognition is crucial for practical uses such as security systems, human–machine interfaces, and video surveillance. When it comes to industrial automation, face recognition systems becoming more and more important. The Deep learning-based techniques performed the greatest in terms of processing, speed, and accuracy for human face recognition. They can manage entry in restricted areas by conforming to the identities of individuals and issuing immediate notification for any security violations. The deep learning method which is mostly used for image recognition is known as CNN. CNN is trained on massive datasets to identify faces accurately in a wide range of situations including varying degrees of illumination, orientation, and occlusion. This paper highlights the introduction of Convolutional Neural Networks, CNN models, data sets, research focuses, and prospects for enhancing CNN-based facial recognition and is beneficial for researchers and scholars for future work in CNN. Research also discusses the role of image compression and noise removal in reducing space and time consumption. Moreover, the role of image quality enhancement using noise removal has been discussed. The research paper also considers the significance of hyperparameters such as learning rate, epoch, batch size, and optimizer in accuracy and performance enhancement.