MSR-Net: Multi-Scale Relighting Network for One-to-One Relighting
This addresses the issue of run-time intensive and memory inefficient methods for photo enhancement through illumination-specific retouching, representing an incremental improvement in the domain of computer vision.
The paper tackles the problem of deep image relighting by proposing MSR-Net, a multi-scale network that aggregates features at different scales to translate illumination from input to target images, achieving high-quality reconstruction through a multi-step training approach with two loss functions.
Deep image relighting allows photo enhancement by illumination-specific retouching without human effort and so it is getting much interest lately. Most of the existing popular methods available for relighting are run-time intensive and memory inefficient. Keeping these issues in mind, we propose the use of Stacked Deep Multi-Scale Hierarchical Network, which aggregates features from each image at different scales. Our solution is differentiable and robust for translating image illumination setting from input image to target image. Additionally, we have also shown that using a multi-step training approach to this problem with two different loss functions can significantly boost performance and can achieve a high quality reconstruction of a relighted image.