Facial Image Super-Resolution with CNN, “A Review”

This survey paper discusses the various applications of super-resolution techniques with their various network designs using deep convolution neural networks. Recent research in state-of-the-art methods has been compared with their performance evaluation matrix, and the image's loss functions have also been discussed to improve image performance. Facial-based work is becoming more popular in super-resolution due to its more practical application. A brief study is described to improve the network's complexity, which helps new researchers gather information for their network designs, different filter sizes, CNN architecture, and standard data sets. Regardless of the advancement as of late, we distinguish a few inadequacies of existing systems and give future research direction towards the arrangement of these open issues.