Single Image Super-Resolution (SISR) is a classical problem of computer vision that aims to obtain a high-resolution (HR) image from a low-resolution (LR) version. In other words, the objective of SISR techniques is to make an image larger without losing details. One of the biggest challenges of SISR is the existence of multiple solutions for the same image. This makes the mapping between the LR space and the HR space unclear. There are three main methods for SISR: interpolation-based, reconstruction-based and learning-based. In this experiment project we propose deep learning techniques that SRResNet network, an evolution technique from standard CNN. 
Experiment project has the objective to validate the match between SRResNet and candidate AI algorithm Self-Supervised Learning (SSL) for Computer Vision to elaborate and predict spatial context, colorization, equivariance to transformations alongside unsupervised techniques such as clustering, generative modeling and exemplar learning. The final results refer to the elaborated image generated through the Single Image Super-Resolution (SISR) techniques mentioned earlier. This technique aims to enhance the resolution of a low-resolution image and produce a high-resolution version with improved details. Along with the image enhancement, there is a partially automated tagging process implemented to identify and label the objects present in the image.