[Boards: 3 / a / aco / adv / an / asp / b / bant / biz / c / can / cgl / ck / cm / co / cock / d / diy / e / fa / fap / fit / fitlit / g / gd / gif / h / hc / his / hm / hr / i / ic / int / jp / k / lgbt / lit / m / mlp / mlpol / mo / mtv / mu / n / news / o / out / outsoc / p / po / pol / qa / qst / r / r9k / s / s4s / sci / soc / sp / spa / t / tg / toy / trash / trv / tv / u / v / vg / vint / vip / vp / vr / w / wg / wsg / wsr / x / y ] [Search | Extra juicy! | Home]

How can I use facial recognition and write a script to automatically

This is a blue board which means that it's for everybody (Safe For Work content only). If you see any adult content, please report it.

Thread replies: 50
Thread images: 11

File: 1404296154265.png (1MB, 1733x2000px) Image search: [iqdb] [SauceNao] [Google]
1404296154265.png
1MB, 1733x2000px
How can I use facial recognition and write a script to automatically harvest all potential reaction images in an anime?
>>
File: wow anon.jpg (78KB, 884x574px) Image search: [iqdb] [SauceNao] [Google]
wow anon.jpg
78KB, 884x574px
It's already bad enough that people are pausing their anime just to screencap potential reaction pics, now people can't even be bothered to do that?

Just wait for someone else to do it for you.
Or is this an ego thing and you wish to be the very first with fresh OC on /a/?
>>
>>47785848
Take a look at OpenCV
> AWOO !
>>
>>47785848
OP, who is this qt semen demon?
>>
>>47785892
I don't even browse /a/, it's a shithole. I lack reaction pics so it would be cool to automatically get some.
>>
>>47785982
warosu girl
>>
you people are really pathetic
>>
>>47785991
>it's a shithole
shit taste detected

also just take your own
i doubt your script can match your sense of humor in any way shape or form.
>>
>>47785848
Wow, what a great idea.
>literally scanning every frame for anime faces
>scanning upwards of 39,600 frames per episode
Your folder would fill up with tens of thousands of almost-identical anime pics.
>>
>>47786086
so then you would just require a certain amount of change before you took another picture why are you so dumb
>>
>>47786105
How much change tho?
You're still going to be capturing the same anime face every time they flap their mouth once.
>>
>>47785848
Write your own self aware artificial intelligence.
>>
>>47786245
I was hoping that with currently already available tools, like some open facial recognition tool, I would be able to do this without any heavy programming.
>>
>>47786277
>facial recognition tool
>anniemay
>>
>>47786277
Pretty much all facial recognition libraries are geared to detect human faces, not drawings.

You would have to roll your own cute anime girl OCR.
>>
>>47786302
I wonder if there's any money in machine vision that's geared toward anime.
>>
>>47785848
You're kind of a fucking idiot
>>
Your father is ashamed of you.
>>
>>47785991
>it's a shithole
I agree
>>
File: image.jpg (2MB, 685x9952px) Image search: [iqdb] [SauceNao] [Google]
image.jpg
2MB, 685x9952px
>>47785848
Just pause the fucking anime to take a reaction image

Holy shit OP how stupid are you?
>>
>>47786469
What's wrong with wanting to automate something?
>>
>>47785848
I've tried this. Existing classifiers are trained on real human faces in photos and cannot recognize drawn characters. You would have to train your own. Do you have a dataset of a couple hundred thousand anime faces?
>>
>>47786541
No, but I have a feeling if I save enough images I might be able to. Of course 200,000 is a lot so I would need to automate it. Now I'm stuck in a loop.
>>
https://www.youtube.com/watch?v=yBACqpTEzT8
>>
>>47785892
people pause their anime to take screenshots? I just mash F5 whenever I see anything happening

>old screenshots folder several GB's strong
>>
>>47786747
I often go back and pause through a scene if I realize it would be a good reaction image later. If I try to anticipate then I take myself out of watching.
>>
>>47786527
Doing automation means doing something productive and not being a complete life-long failure.
>>
>>47787642
Alright then I'll keep doing it. I don't get what your problem is.
>>
>>47786130
You could create a db of the eigenvalues of already added faces to avoid that problem.
>>
>>47787707
You're not supposed to try to do productive things on /g/. If you are more ambitious then just wanting to set up different flavors of Linux all day then you don't belong here. GTFO back to /diy/.
>>
File: 1430342174570.jpg (43KB, 245x279px) Image search: [iqdb] [SauceNao] [Google]
1430342174570.jpg
43KB, 245x279px
>>47785848
I want to ravish Momiji doggy style while she howls.
>>
>2015 year of your lord Jesus in the sky
>not embracing digital minimalism
>not deleting all of your hoarded trash
>not experiencing that feeling of relief as you know there are no pictures or webms to distract you from browsing, working, watching series, shitposting or gaming
>not realizing people don't really care about your face when tfw
>>
>>47787806
>nothing to distract you from watching anime and playing games
>no trash
oh boy, you're such a good minimalist
>>
>>47785982
momiji from touhou.
>>
>>47787823
>anime
Anime sure is trash, but I didn't mention it in the post you're quoting. Nothing wrong with games though.
>>
>>47786469
OMG those /pol/ comics are absolutely hilarious.
>>
Indians would be better suited for that job, OP.
>>
>>47787847
>Nothing wrong with games though.
kek
>>
>>47785848
>>47785982
>>47786000
>>47787834
Awoooooooooooo
>>
Train a haarcascade XML using a lot of anime gurlz' faces as positive images. Then use it on Viola-Jones algorithm. OpenCV should do it. :)
>>
File: 1408446766778.png (639KB, 716x1087px) Image search: [iqdb] [SauceNao] [Google]
1408446766778.png
639KB, 716x1087px
>>47789832
>>47789821
>>47789810
>>47789797
>>47789787
>>
File: 1429674708478.jpg (38KB, 362x346px) Image search: [iqdb] [SauceNao] [Google]
1429674708478.jpg
38KB, 362x346px
>>47789849
>>47789832
>>47789821
>>47789810
>>47789797
>>47789787
What the fuck?
>>
Would /a/ be the best place to gather images to train the anime face recognizer? I think it becomes redundant if I gather a bunch of reaction images to train one, because then I already have a ton of images.
>>
>>47786541
holy shit you tried it?
>>47786566
not only do you have to save enough images you will probably have to label where the face is.

Or you could throw deep-learning at the problem and see what sticks. Deep learning is all the rage right now.

AWOOOOOOOOOOO!
>>
>>47785848
ESFORES, PLS

>>47789787
>>47789797
>>47789810
>>47789821
>>47789832
>>47789849
ITS NOT FRIDAY FAGGOTS
>>
>>47793261
wat the fuck happened
>>
ITT Privileged racist white racists with nothing else to do
>>
>>47786747
>several
>less than 20GB
>>
>>47785892
>ego
>anonymous imageboard
>>
>>47786469
Obesity has actually been found to be a protective factor against some conditions. So has smoking. Their cons still far outweigh any protection.
Thread posts: 50
Thread images: 11


[Boards: 3 / a / aco / adv / an / asp / b / bant / biz / c / can / cgl / ck / cm / co / cock / d / diy / e / fa / fap / fit / fitlit / g / gd / gif / h / hc / his / hm / hr / i / ic / int / jp / k / lgbt / lit / m / mlp / mlpol / mo / mtv / mu / n / news / o / out / outsoc / p / po / pol / qa / qst / r / r9k / s / s4s / sci / soc / sp / spa / t / tg / toy / trash / trv / tv / u / v / vg / vint / vip / vp / vr / w / wg / wsg / wsr / x / y] [Search | Top | Home]
Please support this website by donating Bitcoins to 16mKtbZiwW52BLkibtCr8jUg2KVUMTxVQ5
If a post contains copyrighted or illegal content, please click on that post's [Report] button and fill out a post removal request
All trademarks and copyrights on this page are owned by their respective parties. Images uploaded are the responsibility of the Poster. Comments are owned by the Poster.
This is a 4chan archive - all of the content originated from that site. This means that 4Archive shows an archive of their content. If you need information for a Poster - contact them.