Development of image super-resolution framework

There are some scenarios where the images taken of low resolution and it is hard to judge features from them, resulting in need for enhancement. Super-resolution a technique produce high-resolution image lower-resolution image. The intention here develop system that enhances faces satellite by integrating these models providing an interface access this model. have been various ways achieving super-resolution using different techniques. Throughout years, techniques involving deep learning methods, interpolation techniques, recursive networks explored. We find promising use generative adversarial (GANs). has deployed through Google Collaborate, Python libraries, TensorFlow framework. To assess developed system, which consists images, three metrics calculated. namely, peak signal-to-noise ratio, mean squared error, structural similarity index. model successfully demonstrated capability GANs efficiently generating low-resolution given cases. would then be run on standalone server free Internet users facial images.