TE
TechEcho
Home24h TopNewestBestAskShowJobs
GitHubTwitter
Home

TechEcho

A tech news platform built with Next.js, providing global tech news and discussions.

GitHubTwitter

Home

HomeNewestBestAskShowJobs

Resources

HackerNews APIOriginal HackerNewsNext.js

© 2025 TechEcho. All rights reserved.

CosAE: Learnable Fourier Series for Image Restoration

69 pointsby E-Reverance23 days ago

8 comments

maxbond23 days ago
I've been dabbling in using Fourier analysis in deep learning lately, and I'm surprised it that I haven't turned up very much research in this area (Fourier Neural Operators being what seems to be the biggest exception). Fourier analysis is such a ubiquitous tool, intuitively I'd think it would work great for deep learning. My suspicion has been that complex numbers are difficult to work with, and maybe I'm just bad at surfacing the relevant research, but I'd be interested to hear from those better informed. (My naive approach has been to simply concatenate the real and complex components together into an n+1 dimensional tensor, but surely there's a way that better respects the structure of complex numbers.)
评论 #43814166 未加载
评论 #43809095 未加载
评论 #43812529 未加载
nullc23 days ago
Might be useful to use gabor filters as the basis function, since just 2d cosine filters doesn't produce particularly sparse output for angled features. The additional overcompleteness would probably be helpful for the NN learning.
评论 #43809430 未加载
sorenjan23 days ago
These results look incredible, and with an inference time of only 36 ms for a 4X super resolution on a V100.
评论 #43807863 未加载
syockit23 days ago
I don't know why but I get this uncanny feeling when looking at the restored images. Maybe it's because I know it is restored, I wonder if I'd feel the same way if I find it in the wilds.
gitroom23 days ago
Been messing with this stuff too so I get the struggle. Cool results but man, waiting on code drops always drives me nuts.
PaulRobinson23 days ago
Wait, all my eye-rolling at the TV&#x2F;film trope of &quot;Computer, Enhance!&quot; de-blurring is now redundant, and that stuff <i>is real</i>?!<p>This looks incredibly impressive as a result, but I&#x27;m wary of the use of metrics like FID to evaluate performance. I can take a high-res image, downsample it, then use the method and measure performance very easily: what percentage of pixels were correctly restored? Instead they&#x27;re using metrics like FID which - while useful for purely generative techniques - seem a little vague for this purpose.
评论 #43815985 未加载
doctorpangloss23 days ago
Autoencoders are catching up. Next: luminosity separated from color and UCS.
dingdingdang23 days ago
No code has been released though?
评论 #43807512 未加载