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

Facebook open-sources Detectron

research.fb.com

Post reply on HN
Print view View original

111–120 of 188 posts

Re: Facebook open-sources Detectron

#111

Post by lovelearning »

ajtulloch wrote: post #65
haeffin wrote: post #56

Earlier quoted context omitted.

It's a bit disappointing, when caffe2 was released it was stated that mobile is a big focus, but things like this don't support mobile (even though some of this was demoed by FB on a phone).

FWIW we have some similar R-CNN models using this codebase running in the FB apps on iOS/Android with Caffe2 - with some neat bells and whistles (eg full execution of the compute graph on the device GPU where available via Metal/OpenGL). I’ll look into adding a mobile tutorial for Detectron if you’re interested.

Not the OP, but +1 from me for such a mobile tutorial. And thank you in advance!

Re: Facebook open-sources Detectron

#112

Post by moreless »

zer0t3ch wrote: post #106
kurtisc wrote: post #81

Earlier quoted context omitted.

For the car captchas, I've found actually clicking all the boxes with part of a car will always be a wrong answer (distinct from when it just makes you answer twice). Instead, you have to click on the squares that you know it thinks are cars. This creates a twisted Turing test situation where, to prove you are a human, you have to pretend to be a machine's idea of what a human is.

Isn't it supposed to learn from you? IE, answering technically correctly but what it thinks is incorrect is slightly annoying for you, but better for the system in the long run. (ie better for everyone)

Everyone? Or Google? I don't feel particularly happy about being used as a lab rat, so no, thank you - I don't care about the quality of Google's AI. If anything, I would purposefully mislead it if I knew how.

Re: Facebook open-sources Detectron

#113

Post by aswanson »

kurtisc wrote: post #81
megous wrote: post #7

So is this the end of Google captchas asking for where the car/sign/whatever is? Will there be a final battle of AIs, where they will kill each other, and the unfettered access to websites over VPN/tor wins and laughs the last laugh?

For the car captchas, I've found actually clicking all the boxes with part of a car will always be a wrong answer (distinct from when it just makes you answer twice). Instead, you have to click on the squares that you know it thinks are cars. This creates a twisted Turing test situation where, to prove you are a human, you have to pretend to be a machine's idea of what a human is.

Twisted Turing test. There's a novel waiting to be written about this.

Re: Facebook open-sources Detectron

#114

Post by linkmotif »

stmw wrote: post #48
linkmotif wrote: post #31

Earlier quoted context omitted.

For some reason, people are offended by gross misunderstandings. This framework is in Python, but it’s a Python binding that sets up code that runs natively (not even sure the details myself; others are writing CUDA). TensorFlow is the same way. It’s in Python, but the computations are not in Python. As you point out, that wouldn’t make sense.

Having written Python -> C bindings before, I am quite aware of how this works... It still has considerable overheads.

But these aren’t those kind of binding. Here the Python just sets things up, and that’s the end of Python. That’s my understanding.

Re: Facebook open-sources Detectron

#115

Post by amrrs »

moreless wrote: post #112
zer0t3ch wrote: post #106

Earlier quoted context omitted.

Isn't it supposed to learn from you? IE, answering technically correctly but what it thinks is incorrect is slightly annoying for you, but better for the system in the long run. (ie better for everyone)

Everyone? Or Google? I don't feel particularly happy about being used as a lab rat, so no, thank you - I don't care about the quality of Google's AI. If anything, I would purposefully mislead it if I knew how.

Don't you think that's how a lot of training data can be generated efficient for future ML/AI breakthroughs?

Re: Facebook open-sources Detectron

#116

Post by yorwba »

Jaruzel wrote: post #79
jraph wrote: post #55

Earlier quoted context omitted.

Has someone ever tried to submit images from Google captchas to Google Images? An answer like "This is definitely a sign" from Google Images would be funny.

I just tried it: http://www.jaruzel.com/files/streetsign.jpg :)

Of course if you start googling for street signs a lot, they'll hit you with a captcha for each search, so that you can't ask Google to solve their captchas for you until you solve their captchas.

Re: Facebook open-sources Detectron

#117

Post by Sir_Cmpwn »

Jaruzel wrote: post #85
nannal wrote: post #84

Earlier quoted context omitted.

Your SSL cert is for the wrong domain.

Your https-everywhere is doing it wrong. I don't have an SSL on my personal site (no requirement to). I do have SSL on other domains that share that IP however.

The idea that you have no requirement to is wrong - your site may not have sensitive information but without SSL it can be MITM'd and used as a vector for malware, ad injection, etc.

Re: Facebook open-sources Detectron

#118

Post by 0xTJ »

kurtisc wrote: post #81
megous wrote: post #7

So is this the end of Google captchas asking for where the car/sign/whatever is? Will there be a final battle of AIs, where they will kill each other, and the unfettered access to websites over VPN/tor wins and laughs the last laugh?

For the car captchas, I've found actually clicking all the boxes with part of a car will always be a wrong answer (distinct from when it just makes you answer twice). Instead, you have to click on the squares that you know it thinks are cars. This creates a twisted Turing test situation where, to prove you are a human, you have to pretend to be a machine's idea of what a human is.

Does it tell you that you're wrong, or simply give you another set. If it is just giving you another set, it may be that it thinks that you can provide more useful data.

Re: Facebook open-sources Detectron

#119

Post by kbart »

amrrs wrote: post #115
moreless wrote: post #112

Earlier quoted context omitted.

Everyone? Or Google? I don't feel particularly happy about being used as a lab rat, so no, thank you - I don't care about the quality of Google's AI. If anything, I would purposefully mislead it if I knew how.

Don't you think that's how a lot of training data can be generated efficient for future ML/AI breakthroughs?

So we will get even more "targeted ads" in our faces? No, thanks. I think ML/AI has a great potential (especially in medicine), but I just don't trust ad companies to use it for any good cause.
Post reply on HN