-
[pdf]
[supp]
[arXiv]
[bibtex]@InProceedings{Liu_2023_ICCV, author = {Liu, Yilin and Li, Jiang and Pang, Yunkui and Nie, Dong and Yap, Pew-Thian}, title = {The Devil is in the Upsampling: Architectural Decisions Made Simpler for Denoising with Deep Image Prior}, booktitle = {Proceedings of the IEEE/CVF International Conference on Computer Vision (ICCV)}, month = {October}, year = {2023}, pages = {12408-12417} }
The Devil is in the Upsampling: Architectural Decisions Made Simpler for Denoising with Deep Image Prior
Abstract
Deep Image Prior (DIP) shows that some network architectures inherently tend towards generating smooth images while resisting noise, a phenomenon known as spectral bias. Image denoising is a natural application of this property. Although denoising with DIP mitigates the need for large training sets, two often intertwined practical challenges need to be overcome: architectural design and noise fitting. Existing methods either handcraft or search for suitable architectures from a vast design space, due to the limited understanding of how architectural choices affect the denoising outcome. In this study, we demonstrate from a frequency perspective that unlearnt upsampling is the main driving force behind the denoising phenomenon with DIP. This finding leads to straightforward strategies for identifying a suitable architecture for every image without laborious search. Extensive experiments show that the estimated architectures achieve superior denoising results than existing methods with up to 95% fewer parameters. Thanks to this under-parameterization, the resulting architectures are less prone to noise-fitting.
Related Material