Hello everyone I am the author of this article Image super-resolution is one of the most popular generative algorithm A neural network takes a low resolution

Hello everyone I am the author of this article Image super-resolution is one of the most popular generative algorithm A neural network takes a low resolution image and has to imagine & generate all the finer details A lot of rapid progress has been made in this field coming from early stage ML models to recent TECOGAN In this article I cover the task of super-resolution, the taxonomy of the algorithm developments, loss functions, performance metrics and popular datasets In addition I have provided detailed code of ESPCN architecture to train a model from scratch and corresponding Paperspace gradient notebook Also attached a pre-trained ESPCN model to play out with Would be happy to discuss any further questions related to this topic