Back-Projection Pipeline

Pablo Navarrete Michelini, Hanwen Liu, Yunhua Lu, Xingqun Jiang

Abstract: We propose a simple extension of residual networks that works simultaneously in multiple resolutions for the problem of image super-resolution. Our network design is inspired by the iterative back-projection algorithm and seeks the more difficult task of learning how to enhance images. Compared to similar approaches, we propose a novel solution to make back-projections run in multiple resolutions by using a data pipeline workflow. Features are updated at multiple scales in each layer of the network. The update dynamic through these layers includes interactions between different resolutions in a way that is causal in scale, and it is represented by a system of ODEs, as opposed to a single ODE in the case of ResNets.