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.

A Face Recognition Algorithm That Outperforms Humans?

108 pointsby gagzillaabout 11 years ago

9 comments

apuabout 11 years ago
First, my earlier comments on a &quot;competing&quot; approach from Facebook may help give relevant context for how to think about these numbers: <a href="https://news.ycombinator.com/item?id=7393378" rel="nofollow">https:&#x2F;&#x2F;news.ycombinator.com&#x2F;item?id=7393378</a><p>Briefly skimming through this paper, it appears that these numbers are not a fair comparison, as this paper uses the <i>un</i>restricted protocol of LFW[1], whereas the other methods in the ROC curve shown in the paper are using the restricted protocol. As you might imagine, the latter is more restrictive -- specifically in terms of amount of training data allowed. And as I mentioned in my previous comment, training data is king in these kind of systems -- more is always better.<p>To go slightly out on a limb, I think more significant than the new theoretical model proposed in this paper is probably the use of lots of different types of datasets for training. (Significantly more data &gt;&gt; more complicated models, most of the time.) But I&#x27;d have to read the paper much more carefully to be sure about this.<p>[1] <a href="http://vis-www.cs.umass.edu/lfw/results.html" rel="nofollow">http:&#x2F;&#x2F;vis-www.cs.umass.edu&#x2F;lfw&#x2F;results.html</a>
评论 #7638165 未加载
评论 #7638268 未加载
dangabout 11 years ago
The original url [1] was blogspam—that is, it was a knock-off (or excerpt) of some other, more original source. In such cases HN strongly prefers the original source.<p>Submitters: blogspam is usually easy to recognize. Please check for that and post the original instead.<p>1. <a href="http://news.sciencemag.org/signal-noise/2014/04/face-recognition-algorithm-finally-beats-humans" rel="nofollow">http:&#x2F;&#x2F;news.sciencemag.org&#x2F;signal-noise&#x2F;2014&#x2F;04&#x2F;face-recogni...</a>
评论 #7637582 未加载
评论 #7637624 未加载
ihodesabout 11 years ago
The actual paper (parts are accessible &amp; interesting): <a href="http://arxiv.org/pdf/1404.3840v1.pdf" rel="nofollow">http:&#x2F;&#x2F;arxiv.org&#x2F;pdf&#x2F;1404.3840v1.pdf</a>
评论 #7638934 未加载
评论 #7638180 未加载
netcanabout 11 years ago
Face recognition is one of those technologies that&#x27;s seems neat at a glance and mindbogglingly terrifying on closer inspection. It has the potential to sci-fi the world overnight and it could do it tomorrow night. The algorithm accuracy and enormous comparison DBs are already here.<p>The effect this can have on commerce, advertising, policing, crime, culture, or a bunch of other things has enough wide reaching effects for a sci fi thriller.<p>A camera in cahoots with a till in a supermarket could put a face and a name on every purchase. If the camera and the till in cahoots with an advertising billboard in a shopping mall, you have created an offline version of conversion tracking.<p>Since the supermarket and billboard company are in cahoots, they can compare notes and find a billboard location that gets the supermarket&#x27;s best customers. If you are seen checking out climbing gear by a camera in cahoots with Facebook, that store can keep outdoor activity products to you in Facebook. Hello offline retargeting.<p>That&#x27;s just advertising. Imagine policing. Imagine high school.
评论 #7639810 未加载
gcrabout 11 years ago
I&#x27;m a computer vision grad student. A few things concern me about this work. Maybe they&#x27;re incidental, but I&#x27;m not ready to throw my hands up in the air quite yet.<p>- Why wasn&#x27;t this accepted to CVPR&#x2F;ECCV&#x2F;one of the well-established computer vision conferences? I would love to read some of the reviewers&#x27; comments about this work before I give further judgment. (If this really is some CVPR preprint, or if it actually is peer-reviewed, I&#x27;d feel much better about this.)<p>- Why isn&#x27;t this work listed on the official curated &quot;LFW Results&quot; page that Erik Learned-Miller maintains? <a href="http://vis-www.cs.umass.edu/lfw/results.html" rel="nofollow">http:&#x2F;&#x2F;vis-www.cs.umass.edu&#x2F;lfw&#x2F;results.html</a> Is this work so new that Erik hasn&#x27;t had time to review it yet?<p>- Human performance on LFW is 99.2%, which is higher than what the authors think it is. The performance drops to the (claimed) 97% when we only show humans a tight crop of the face: <a href="http://www1.cs.columbia.edu/CAVE/publications/pdfs/Kumar_ICCV09.pdf" rel="nofollow">http:&#x2F;&#x2F;www1.cs.columbia.edu&#x2F;CAVE&#x2F;publications&#x2F;pdfs&#x2F;Kumar_ICC...</a> They discuss this difference in a paragraph in their conclusion, but I consider it dishonest to use the lower number in the abstract and imply it in the title. In fact, I consider it misleading to put &quot;Surpassing human performance&quot; in the title to begin with, but that&#x27;s another matter :)<p>- Showing good performance on one dataset (LFW) is certainly not enough to show that this &quot;outperforms humans&quot; in the general case. Getting a state-of-the-art result on LFW these days is like squeezing a drop of water out of a rock; in my opinion, we should turn our attention to harder datasets like GBU now that these &quot;easier&quot; ones are solved.<p>I&#x27;m not terribly familiar with Gaussian processes so I&#x27;m not sure whether the math works out, but it is a pretty uncommon thing to try in this domain. (Perhaps that&#x27;s what makes this work interesting, especially since this year seems to be the &quot;Deep Learning is Eating Everyone&#x27;s Lunch&quot; year)<p>I also wish they describe what final-stage classifier they use for the &quot;GaussianFace as Feature Extractor&quot; model. Often, that&#x27;s the most important step; it&#x27;s strange that they didn&#x27;t compare with POOF&#x2F;High-dimensional-LBP&#x2F;Face++&#x27;s deep-learned features&#x2F;any of the other state-of-the-art feature extractors, especially considering how much worse &quot;GaussianFace as a binary classifier&quot; does (93% vs 97% is a huge difference in this dataset)<p>Just my two cents. It definitely demands further exploration. I don&#x27;t see any obvious mistakes, but I&#x27;m not sure why their approach works as well as they claim it does either.<p><i>Edit</i>: I don&#x27;t mean to start a witch hunt or anything, but if the authors have the guts to put &quot;Human-level performance&quot; in their title, they&#x27;re just <i>begging</i> for the community to inspect every detail and point out all the flaws in every minutiae in their work. It&#x27;s our community&#x27;s hot button. It&#x27;s similar to the old adage about how if you want a Linux user to help you, you have to tell them how much Linux sucks. That&#x27;s where much of my skepticism comes from. The most astounding papers are often the most humble, but &quot;humble&quot; certainly doesn&#x27;t describe this work.
评论 #7638269 未加载
评论 #7640493 未加载
评论 #7638447 未加载
评论 #7638258 未加载
评论 #7638808 未加载
评论 #7643664 未加载
评论 #7638257 未加载
评论 #7638615 未加载
schrodingersCatabout 11 years ago
I&#x27;ll have to run this by my friend who writes morphometrics algorithms, as I can;t actally tell what is new about this paper. This might actually allow for a proper photo-matching search engine. All the ones that I have tried to this point have been lacking or broken...
dschiptsovabout 11 years ago
No matter angle and illumination? Come on.)
评论 #7638211 未加载
michaelochurchabout 11 years ago
Eigen tell you, it&#x27;s not easy.
weishigonameabout 11 years ago
Thanks for sharing.