Fast Fourier Intrinsic Network
This addresses intrinsic image decomposition for computer vision applications, with incremental improvements in efficiency and performance.
The paper tackles the problem of decomposing images into albedo and shading by proposing FFI-Net, a lightweight network that operates in the spectral domain with a novel spectral loss, achieving state-of-the-art performance on MPI-Sintel, MIT Intrinsic, and IIW datasets.
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.