Practical Deep Raw Image Denoising on Mobile Devices: 2020 CVPR Paper

This article presents a practical deep raw image denoising method designed for mobile devices. The method utilizes a novel deep neural network architecture, called 'RAWNet', which effectively removes noise from raw images while preserving details and textures. It also includes a fast noise estimation method that can estimate the noise level of images in real-time on mobile devices. Experimental results demonstrate its effectiveness in denoising under various noise conditions, and its ability to operate in real-time on mobile devices.