Focal Frequency Loss for Image Reconstruction and Synthesis

By Liming Jiang, Bo Dai, Wayne Wu, Chen Change Loy. IEEE International Conference on Computer Vision, 2020

467 citations recorded by the source corpus.

Image reconstruction and synthesis have witnessed remarkable progress thanks to the development of generative models. Nonetheless, gaps could still exist between the real and generated images, especially in the frequency domain. In this study, we show that narrowing gaps in the frequency domain can ameliorate image reconstruction and synthesis quality further. We propose a novel focal frequency loss, which allows a model to adaptively focus on frequency components that are hard to synthesize by down-weighting the easy ones. This objective function is complementary to existing spatial losses, offering great impedance against the loss of important frequency information due to the inherent bias of neural networks. We demonstrate the versatility and effectiveness of focal frequency loss to improve popular models, such as VAE, pix2pix, and SPADE, in both perceptual quality and quantitative performance. We further show its potential on StyleGAN2. 1, 2

Subfields: Computer Vision and Pattern Recognition

View Paper: https://www.semanticscholar.org/paper/05afe403fd398f774aed4e9872a0eca1dfa21d36

How researchers see this paper

The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.

In Computer Vision and Pattern Recognition

Citations in the corpus

MAXIM: Multi-Axis MLP for Image Processing (2022, background): “Lfreq is the frequency reconstruction loss that enforces high-frequency details [19, 34]:”
Camouflaged Object Detection with Feature Decomposition and Edge Reconstruction (2023, methodology): “Deep wavelet decomposition is an effective tool to decompose image/feature into various frequency components and has gained immense popularity in many domains, such as image restoration [15]”