Hacker News Board Hacker News, browsed like a classic message board
Live data from Hacker News

Is It a Duck or a Rabbit? For Google Cloud Vision, Depends on Image Rotation

reddit.com

Post reply on HN
Print view View original

51–60 of 98 posts

Re: Is It a Duck or a Rabbit? For Google Cloud Vision, Depends on Image Rotation

#51

Post by jasode »

stared wrote: post #50
IanCal wrote: post #39

Earlier quoted context omitted.

I'm a little confused, it seems quite accurate to me. It's the famous duck/rabbit rotation illusion and the google cloud vision API returns different results depending on rotation. What do you find misleading about it?

See e.g. https://towardsdatascience.com/breaking-neural-networks-with...

I wasn't the one who downvoted you but I agree with IanCal and this blog title didn't trigger my "clickbait" sensor.

I disagree that this is an example of "adversarial attack". The famous duck/rabbit illusion has been around since ~1892[1] and therefore was not deliberately constructed to be an "adversary" to image classification neural networks.

To me, it's an interesting example of feeding a well-known optical illusion to an AI algorithm and observing its behavior.

[1] https://en.wikipedia.org/wiki/Rabbit%E2%80%93duck_illusion

Re: Is It a Duck or a Rabbit? For Google Cloud Vision, Depends on Image Rotation

#52

Post by mannykannot »

IanCal wrote: post #39
wodenokoto wrote: post #36

Earlier quoted context omitted.

Clickbaity? The title is downright misleading.

I'm a little confused, it seems quite accurate to me. It's the famous duck/rabbit rotation illusion and the google cloud vision API returns different results depending on rotation. What do you find misleading about it?

It is an example of an adversarial title - one that is sometimes mistaken for clickbait even though it is an accurate summary of the content. This response can be primed by repeated prior training on real clickbait.

Re: Is It a Duck or a Rabbit? For Google Cloud Vision, Depends on Image Rotation

#54

Post by mannykannot »

FakeComments wrote: post #42
SiempreViernes wrote: post #34

Earlier quoted context omitted.

I think this shows how poorly many neural networks are at handling ambiguity. The picture is constructed to be ambiguous, and this property is preserved by the rotation: you can still easily see the duck by slightly shifting where you focus. One mode might be more prominent at some orientation, but the ambiguity is always there so to confidently assign labels and then switch what you assign is an error. So you should…

The neural network is likely handling it just fine: A classifier generally outputs a vector of weights, so it’s likely classifying, say [0.8, 0.75] and then the output is selecting the highest and saying “bunny”. Then you rotate it, and the classifier says [0.75, 0.8] and the output says “duck”. This is completely reasonable on the part of the network: all things being equal, animals generally appear in certain orien…

It works that way for me when I play it. To see this, it seems to help to look at the picture but have half an eye on the changing NN evaluation. I don't think my perception is simply responding to the changing NN output, as there are times when I disagree with it. My guess is that, by dividing my attention, I reevaluate the picture more frequently.

Re: Is It a Duck or a Rabbit? For Google Cloud Vision, Depends on Image Rotation

#55

Post by stared »

jasode wrote: post #51
stared wrote: post #50

Earlier quoted context omitted.

See e.g. https://towardsdatascience.com/breaking-neural-networks-with...

I wasn't the one who downvoted you but I agree with IanCal and this blog title didn't trigger my "clickbait" sensor. I disagree that this is an example of "adversarial attack". The famous duck/rabbit illusion has been around since ~1892[1] and therefore was not deliberately constructed to be an "adversary" to image classification neural networks. To me, it's an interesting example of feeding a well-known optical illu…

Yes, I agree with you. (No idea who downvoted my and why either.)

This (well known) illusion is NOT an adversarial example. Though, I explain why for people working with AI (e.g. me) the title seemed like mentioning an adversarial example. There are plenty of examples of "just rotate and a vulture becomes an orangutan" where it does not look like an orangutan for humans.

Vide: "A Rotation and a Translation Suffice: Fooling CNNs with Simple Transformations" https://arxiv.org/pdf/1712.02779.pdf

Re: Is It a Duck or a Rabbit? For Google Cloud Vision, Depends on Image Rotation

#56

Post by ozzyman700 »

devin wrote: post #24
salty_biscuits wrote: post #22

Earlier quoted context omitted.

But only with context right? It is completely ambiguous if the character is 6 or 9 without some other clue, like which way up the paper is, or from the 4 next to it (if you assume that an arbitrary rotation may have occurred). It is just a sign to me that doing some rotations as data augmentation is not good enough. Rotation invariance needs to be built into the architecture of the network (like translation invarianc…

How often do you read upside down?

every meeting

Re: Is It a Duck or a Rabbit? For Google Cloud Vision, Depends on Image Rotation

#59

Post by minimaxir »

minimaxir wrote: post #3

Creator of the animation here. Most of the relevant information/context behind the animation (including a link to the repo) is in this Reddit comment: https://reddit.com/r/dataisbeautiful/comments/aydqig/_/ehzyo...

To answer the question why I made the animation: there isn't an ulterior "I found an AI gotcha!" motive, I saw a tweet where the API returned different things depending on orientation and expanded on it. It was also an opportunity to test a few animation hypotheses via gganimate.

Re: Is It a Duck or a Rabbit? For Google Cloud Vision, Depends on Image Rotation

#60

Post by gambler »

ratel wrote: post #46

I quite surprised at the comments on HN so far as nobody seems to see the significance of this. Yes, the image is ambiguous. The point is that Google Cloud Vision gives an unambiguous answer of that image based on the rotation. Transformations of an image are regularly used to improve the results of image recognition. That process fails quit dramatically if in the course of a transformation the answer given is presen…

I'm glad that at least someone here sees the problem, but I am not surprised by the typical reaction of AI apologists in this thread. You always get at least one of the two responses:

"OMG, this is amazing, it's just like humans. We're probably close to AGI."

"Ha-ha, humans are stupid, so the algorithm giving unexpected result is just a proof that it's better and less biased."

Here, we have both in response to the same demo.

Still, I honestly don't know why some people are so biased in favor of neural nets and have zero interest in edge cases and flaws (the most interesting parts if you want to gain deeper understanding of how the algorithm actually operates). Wishful thinking, I guess.

Post reply on HN