Fast Fourier Intrinsic Network

Yanlin Qian, Miaojing Shi, Joni-Kristian Kamarainen, Jiri Matas; Proceedings of the IEEE/CVF Winter Conference on Applications of Computer Vision (WACV), 2021, pp. 3169-3178

Abstract

We address the problem of decomposing an image into albedo and shading. We propose the Fast Fourier Intrinsic Network, FFI-Net in short, that operates in the spectral domain, splitting the input into several spectral bands. Weights in FFI-Net are optimized in the spectral domain, allowing faster convergence to a lower error. FFI-Net is lightweight and does not need auxiliary networks for training. The network is trained end-to-end with a novel spectral loss which measures the global distance between the network prediction and corresponding ground truth. FFI-Net achieves state-of-the-art performance on MPI-Sintel, MIT Intrinsic, and IIW datasets.

Related Material

[pdf] [supp] [arXiv]
[bibtex]
@InProceedings{Qian_2021_WACV, author = {Qian, Yanlin and Shi, Miaojing and Kamarainen, Joni-Kristian and Matas, Jiri}, title = {Fast Fourier Intrinsic Network}, booktitle = {Proceedings of the IEEE/CVF Winter Conference on Applications of Computer Vision (WACV)}, month = {January}, year = {2021}, pages = {3169-3178} }