Supervisors: Dr. Zhiwu Huang, Dr. Danda Pani Paudel, Prof. Luc Van Gool
Image enhancement is a long standing problem in the field of computer vision. In this work, we investigate how to enhance high resolution images, using generative adversarial network, in the setting of unpaired images. Our approach comprises two main aspects: (i) multisliced modeling of the enhancement problem for which a new network architecture is designed and optimized in a divide-and-conqure fashion, (ii) learned bilateral grid based image enhancement for a faster computation. With several experiments, we show that the proposed multisliced technique performs better than the current state-of-the-art methods, for both paired and unpaired image enhancement settings. Furthermore, we also achieve more than 3x speed-up by using the bilateral grid based enhancement, within the multisliced modeling, without compromising the accuracy significantly.