SuperNet: An Efficient Image Super-Resolution Preserving Neural Network

Due to large development in multimedia information processing and fashion styling in commercial and social applications, cloth image recognition creates a huge impact. However, the large variations due to blurring in cloth image appearances and their complicated style formation conditions create challenges in image recognition. Moreover, Super Resolution (SR) technique raises the high frequency components that are used to generate a high-resolution image with good perceptual quality from a low-resolution image. A novel residual deep neural network called SuperNet approach that converts a low-resolution image to a high-resolution image by providing more advanced features for better characterization of clothing genre is introduced in this work. Furthermore, the proposed framework reduces the complexity of the network without content loss of the original image.