Jewdar (tm)

COMPUTER ASSISTED JEW DETECTION

Doesn't that sound nice?
Shalom goyim. I'm experimenting with image classification using neural networks and I suddenly had a thought.
You know how researchers where able to detect homosexuality with AI? I'm certain this technique works on hour hebrew friends

Kikes hate being named more than anything, the AI will not only make it easier, it might allow us to discover well hidden cryptos.

Of course not, the possibilities are endless if the problem involves pattern recognition.

Goals
>Write some software (someone pls halp, i suck at coding) that allows us to make more qualified guesses on (((someones))) jewishness

What I really need right now is image sources, folders, websites, you name it. Gathering 10000 images is no easy task.

Maybe you have other ideas?
Post them here, anything that allows us to further our cause is welcomed.

Other urls found in this thread:

image-net.org/synset?wnid=n09607903
vis-www.cs.umass.edu/lfw/#download
majorityrights.com/uploads/jewish-last-names.txt
ghostbin.com/paste/rpsh7
flickr.com/search/?text=
medicaldaily.com/rare-genetic-disorders-could-be-diagnosed-facial-recognition-computer-software-289688
imdb.com/title/tt1019452/fullcredits?ref_=tt_cl_sm#cast
mega.nz/#!dCRDwYCZ!93IDZaH3crTQ0NmIaNeyWzL_ml-jZ-cI9p_5rUkc7sw
ghostbin.com/paste/avrwv
vid.me/Brother-Nathanael
imdb.com/interfaces/
en.wikipedia.org/wiki/ACTOR
coincidencedetector.com/
mega.nz/#!MJ8kmbYD!T7ahVYln8ysnA5EoXULug5ODKz_DnYCOWUuTCrdtiqE
a.uguu.se/hYKxVW3nmuyb_Jew.rar
archive.fo/VYihQ
archive.fo/Z8e6j)
notabug.org/JohnnyHobo/CommunityPolicing
notabug.org/JerryShaw/EyeOfAnEagle.
smoloko.com/?p=10885
coincidencedetector.com/theList.json
coincidencedetector.com/judasList.json
judas.watch)
coincidencedetector.com/theTree.json
imgur.com/a/09ieS
hooktube.com/watch?v=XOxxPcy5Gr4
cs229.stanford.edu/proj2016/report/ChenDengZhang-WhereAmIFromEastAsianEthnicityClassificationFromFacialRecognition-report.pdf
openu.ac.il/home/hassner/projects/cnn_agegender/CNN_AgeGenderEstimation.pdf
mega.nz/#!1T5iwRTJ!YuQbNoZ7GsLWMWgN3w6pRhkbiRGIwnI_o5Y2m8DP7kg
en.wikipedia.org/wiki/Lists_of_Jews
famousfix.com/list/jewish-celebrities
bakerdh.wordpress.com/2012/05/05/face-distortion-is-not-due-to-lens-distortion/
sci-hub.io/10.1016/j.cognition.2011.08.001
vk.com/dev/users.search
vk.com/search?c[per_page]=40&c[photo]=1&c[religion]=Judaism&c[section]=people
vk.com/search?c[country]=8&c[per_page]=40&c[photo]=1&c[section]=people
dianne.intec.ugent.be/#gettingstarted
dl.sony.com/
pastebin.com/ihT3Nk58
seductivejewess4.net/
twitter.com/NSFWRedditGif

bamp

(dubs of truth)
oy vey goy there is no room for an anti-semitic bot on the internet! theres enough hate speech already!

>>>/gaschamber/
Go back. Have a picture of a gentile incase you are serious.

YOU OWN NOTHING GOYIM

It merely sucked on winkike

Now with script, its dog slow though

image-net.org/synset?wnid=n09607903 - if you click Treemap Visualization, then go up to Jew, Hebrew, Israelite there are a few more categories of kike

Here are 13k images of faces with names - vis-www.cs.umass.edu/lfw/#download - you can filter against majorityrights.com/uploads/jewish-last-names.txt

You can write a script that searches Twitter for kike names, or tweets from Israel, then use OpenCV to check if the profile pic is a face. If it's a face then add it to your Kikédex.

GODDAMN GOOBERGATORS!!

...

Nice! I'll grab the imagenet and umass datasets now, then twitter depending on the amount of images

...

...

...

shame on you

...

So many assblasted kikes itt already.

This will really just end up with 20k pictures of Trump filed under jew

kek'd

Oy Vey! We might need the UN to regulate AI development now!!!


This is a good way to get done things they otherwise would not do, make it anti Jew someway, then suddenly BAM! Everyone is involved.

Got 4 for ya.

I'm downloading the image net data now. Alot of the urls seem dead. Now for the umass dataset, hopefully it is more fruitful.

This is hilarious.

Why not just use (((Coincidence detector))) and datamine the names which leads you to faces?

Your doing Skynets work good sir.

Does it have a public DB of names?
Maybe i can make a facebook/twatter/linkedin crawler

Do you know anything at all about neural networks?

Knowledge doesn't hurt. Imagine if I were to criticize the Bible by mentioning how Buddha married to 6-year-old Aisha. It would be retarded, right?

So before criticizing something, go learn more about it. It's just math. "Neural networks" just means lots and lots of calculus. There's nothing inherently Jewish about matrices and SIMD operations.

I can't wait for when they inevitably make some google glass-type gadget popular, and people start using our 'irl video coincidence detector' app

Trips of Truth

I hope it will be able to run on smartphones, starting with a very small net for that reason. I want to find a good balance between resource usage and precision.

if you need help with your code, there's probably a million tutorials on how to build a neural network out there. I had to learn how to do it last year; it's not that hard once you learn the algorithm.

Could probably cut the work load to a fraction by narrowing down criteria. Instead of having it identify a face and then run through 1000's of comparisons maybe have it do several id's

Dubs confirm.

From the production values I thought that was going to be anti-Christianity propaganda, but it can't be that this is seen as anti-Christian, right?

There is, I just started fiddling with it last week. Maybe i should make a repo on gitgud

The network already does a lot of this, thats the beauty of it. But transforming the face into a more simple kind of data is a good idea

I've been dreaming of this neural network for awhile and want to make it a reality. The only real limit to this is processing power. If you do end up making a repo link it here and I'll make a burner to contribute. Creating True Artificial Intelligence is the one thing we can use to overpower the kike. The answer to the control problem is simple: truth.

Bear in mind that this series was only localized by HBO so there's no kike hands in the making. Show is basically if a Holla Forumsack became Pope. He openly hates fags, pedos and any degenerates. I thought it would be blasphemous at first but was pleasantly surprised. I also memed Trump as the Young Pope to my boomer parents.

...

Regulating research and/or development is useless because it means non-compliant teams will advance a lot faster. At best they could draft up a criteria for what governments/corporations should approve for live public applications. They'd probably call it something like: U.N Artificial Intelligence Stability and Ethics Criteria or AISEC certification for short. Even then it'll have to function as a guideline only, mainly to safeguard against faulty applications being applied to public systems like transit, health, finance and defense.

Inevitably the only real solution though is "better AI beats AI". So they'll just have to devise counter-strategies to build better AIs to mess with AI developed for a purpose against theirs.

Isn't Last Cuck Tonight also HBO?
Meme the Gen Z instead. Go bring them the nice gift they want, and drop micro-redpills like "Do you like how society is? Society doesn't have to be like it is now, long ago someone tried to change it but they failed."
Kids are smarter than people give them credit for.

Well guess who's not complying?

Here are a few, but they are marred by poor typography and kike-stars.

GET THIS RUNNING ON IBM's OPEN SOURCE QUANTUM COMPUTERS FOR MAXIMUM BUTTHURT

I'll agree with the poor typography, but consider pic 4. Without the star, normies look at it and think "oh yeah, it's white people, they dumb xD"

Available to the public or whatever, you know what I mean.

It's fine, i can do some cv2 magic if needed. All data helps!

Checking if I'm still banned, let's see…

Yes it is which is why I was surprised. I already do that as well. Autists seem to be the only ones perceptive enough to accept the cold hard truth however

Great, alright - if I create a collection of thousands of images of kikes where the fuck am I supposed to upload that?

mega, dropbox or any other hosting site.

Yeah, also be extra nice to kids whose parents are shit. They will grow up, and remember you, plus you'll get to feel nice for making a poor kid's day. You can't truly redpill kids but you can sow the seed.

I'm well aware user I was doing that even before I knew the truth. Everyone here should realize being a good role model pays back two-fold. All it takes is treating them like an equally sentient being.

You writing this with python, user?

Mostly, network descriptions are caffes standard files

TFW this project is the killer app for google glass.

Good idea OP though if I may make one suggestion let's change the name to Jude-ar

Surely, there must be some better alternative…

It's perfect

Are we entering the future?

OP here, ready to remove all images from umass dataset that aren't jewish family names

Worked with pepe

Fund it.

Dump that cleaned list into a ghostbin, would ya? Going to take a shot at getting more pics from Twitter.

Sure!

ghostbin.com/paste/rpsh7
Here you go

This project has been blessed by Kek. Do not let it go to waste

Thanks user!

It's also a necessity if Holla Forums intends to stay in the game of information warfare. The applications are potentially endless, especially if Holla Forums can design an autonomous redpilling agent that can analyze a target and determine the best strategy to reach out to them.

Praise kek

Thanks! Got OpenCV up and running in Python, I'll be looking for kikes on Twitter now.

This is fucking datamining when its best, look and weep jewge, kikrosoft and facehook.

Taking a bath, then I'll be back

So for the gentile pictures are you only going to show it whites?

Hi jew

Does it only work in 2.7?

That would imply being a fag is almost completely a non-choice. While studies with twins show that faggotism isn't genetic, if an AI can tell from facial features alone then it can't be due to the way kids are raised or trauma in the vast majority of cases. What does this mean?

Got the DLL error too, eh? Yeah.

maybe other factors such as expression or such. Like the way they put the faggy voice on but with their face muscles.

Yeah configuring rn

We should eventually add classifications, or at least datasets, for other ethic groups. It would be nice for when we make a RWDS, Right Wing Drone Squad.

Isn't OpenCV just a detection librarry without a neural network? You'll need something like TensorFlow.

forgot pic

Doing that could result in muds giving false positives, it needs very diverse data :^) I suppose it would even be possible to make it racially aware, if that is wanted

We are talking about very small variations here, that become statistically significant by sheer sample number. Even if programmed as well as possible today, the software would not be reliable enough in individual cases. I think a trained artist would be a far better judge of physiognomy.

You can use OpenCV as a preprocessing step. For example, extract faces from larger images, resize into desired format, save. Furthermore, OP is using Caffe which is like a TensorFlow alternative.

OP, store the dataset somewhere, I want it as well.

t. Machine Learning PhD

I'll publish results anonymously on ArXiv for maximum buthurt.

If it's possible to detect homosexuality, I certainly think ethnic differences are possible

I'm not writing the Jewdar, I'm trying to help by getting data to train the Jewdar - kike profile pics. Gotta make sure it's not a dog or a bagel or whatever.

Ah, nice that we've got someone serious here.
t. first year CS bachelor

Understood.

Nice having some experts ITT. I will publish everything, looking forward to reaping the salty reward. The reason i used caffe is that it seems like the best alternative if this is supposed to be multi platform, also I'm too poor to buy a new GPU right now.

The study is not as simple as the guardian headline makes it out to be.
Prenatal testosterone exposure is reflected in the facial proportions, but is not a guaranteed determinant of behavior. The same would be true for other factors.

If you can make this with 90% accuracy, we'll be able to organize a white jew-free movement in no time.

if it can detect (((rhinoplasty patients))), doubly effective.

make sure to get some confirmed aryans in there for verification and refinement

To expand on my comment: of course ethnic differences can be detected, but only to a degree. At some point, there is too much overlap for it to be reliable.

Personally, I prefer PyTorch, which provides the best performance on GPUs. But Caffe is pretty good as well. However, I think the PyTorch syntax makes more sense for building the computation graph.

Also, if you have multiple computers laying around I could point you to some papers on distributed optimization to speed up the training (if your model is sufficiently large, and CPU is saturated). I'm an (((expert))) on that topic.

I'm sure someone has a coin mining rig they could spare some time on.

go to a dating website for images of Jews

jewmatch.com or some shit

Front pictures only offer a fraction of the information about the forms of a face. The profile matters a lot as well. Ideally, we would be matching 3D scans to a database of ethnicities and return percentage values for how close a match is. But that would be a multi million dollar project to realize, and even then, would require solid scans of the subjects to begin with.

flickr.com/search/?text=
add jewish related search terms as required. using holiday terms will turn up more group photos.
partial terms work for searches

You have models which estimate the geometry given a flat 2D picture, but I don't think it will supply you with that much information. It will only make the optimization harder. The thing we really need now is a large (((diverse))) dataset. I don't think grabbing pictures from jewmatch.com or something is a good idea, unless you want to build a homosexual pedophile classifier or something.

Twitter search seems to be working.

the gaydar ai had 5 lowres images of each person to detect gayness

medicaldaily.com/rare-genetic-disorders-could-be-diagnosed-facial-recognition-computer-software-289688

software for rare genetic disorders…

still trying to find what software to use? and how to use it resourcefully and without jewgle or judazon cloud services etc.

you can enter the door saying you want to research how to detect illnesses, for medical research and then just use it for jew detection, if you're going to use it as ind. researcher.

whatever is used, if we can discriminate against jews with the software and they keep maintaining themselves as 'white', it can possibly be gotten away with

i'd say we have something to lose doing this, but jews already do this nepotist shit, so there's literally nothing to lose

How consistent were those pictures in terms of lighting, facial expression, camera angle, camera lens, person's age, etc?

I think if you are going to build a database, it needs to be very well curated and selected for subjects that would be the most important to actually be able to tell apart. It would not be very helpful if it was just filled with stereotypes that everyone could tell apart at first glance.

about as curated as a dating website

There are tricks you can use to avoid variance in data. Histogram normalization comes to mind

oy vei

In the worst case scenario if this software doesn't perform as we hope, we should send screencaps of this thread to jewish media to make them freak out and get paranoid.

Holy fuck your last pic/gif is terrifyingly creepy.. spoiler that shit next time

Contributin…

...

I'm looking at too many jews, my eyes are burning.

...

...

Spamming the thread with superjews will not help.

...

jep, a discussion thread with 16 million pictures of jews will not help. you can do that when we have a dedicated place for that

...

Just upload these on mega or some other site, much faster

...

Retard, these classifiers need THOUSANDS of images as input. Spamming the thread with dozens of images isn't helping.

Alright, I'll stop posting Twitter kikes and just post a link when I've caught them all.

thank you kind sir

Shut up, faggot. Every bit helps

are you trying to get this thread off topic?
who is going to download 20.000 pictures manually from pol?

get fucking fucked m8

but seriously, this is awesome

We need to figure out how to leverage GPU/CUDA. So many anons have so much GPU.

requesting quadroon ai

filtered

OP here with update
We currently have 1000 kikes, thanks to all Holla Forumsacks showing interest. Looking forward to your data!

The model should be platform agnostic, as long as your GPU supports OpenCL or CUDA it should be fine, image classification should be doable on CPU

Once this is done, somebody needs to hook this up to a "smart" oven that turns on whenever a camera sees a jew ring the doorbell.

It's a shame more information can't be acquired about AlphaGo Zero. Its algorithms demonstrates a more generalized self-play learning by preset reward functions and basic understanding of the game. Obvious problem is that the processing power required for it is out of Holla Forums's reach for now (if you want results in a satisfactory time frame).

Re-purpose that and you could make it learn how to amplify facial recognition learning, by treating it like a puzzle game of how to draw a target face from learned samples, and evolve generalized algorithms that make it the perfect profiler. Reward mechanism would motivate accuracy and classifying common factors of certain facial types.

In theory it would mean only need a small data sample and basic rules for it to follow. I wouldn't be surprised Facebook eventually approaches Google on this possible collaboration.

...

Maybe a nice idea, but it sounds like you have no clue about nn.

What is your network architecture / what are your computing resources? What is the dimensionality of your input space? Have you done any calculations about how much computing power you'll need, given input dimensionality?

How will you account for different facial orientations and different color profiles in the images?

Have you calculated your vc-dimension to get an estimate on the number of training points you'll need, given your input dimensionality? What did you get?

Ive realized every ugly white person on tv Ive seen since my youth is a jew

unfortunately I am not a coder but when there is anything i can do to help, let me know.
I am speaking for myself and many others with similar autism.

Currently at 840 profile pics from Twitter, I'd say 95% are obviously the kikes the account belongs to and 5% are pics of other random shit that has a face.

Would it be feasible to integrate this into a distributed computing project like SETI@home? JEWDAR@home

apache hadoop

I am by no means an expert, this is a learning experience for me as well.

Noice, i will begin gathering gentile images as well

Think back to all those headlines over the years – how they tried to intimidate us with 'racist electrolysis' and 'fascist face detection' and many other Zuckerberg/DHS/ZOG projects from a future Jewish dystopian hell.

Then think about the last 1 year or so – their kvetchy reactions to our memes.

Now imagine if we come to the table with software that is building a massive neural net to sniff them out, automatically, 24/7. Imagine the absolutely neurotic reaction they will have if they ever see a headline about this. It's time for them to be scared to death and lose sleep …

This is cyberpunk.

Posting a few more of these cursed creatures. Up to 1200 now.

Could classifying a face as jewish be used as a proof of work function?

Hey faggot, you just feed the net the data and it poops out results. You don't need to worry about shading and angles and shit. They learn better when you don't force feed them the lessons.

Maybe someone can share their scraping script

But i have worked with CNN's before, made a classifier that could differentiate real/cartoon images with 98% accuracy. This one will be definitely be more challenging, but facial geometry classification networks already. To account for color/light difference, I will be using histogram equalization, different facial orientations can be detected in cv2 and corrected/deleted.

great, I am currently gathering gentiles, should have a couple thousand in a bit.

12K GENTILES IN STOCK
ONLY KIKES ARE NEEDED

For the first experiment I think jew/not jew is the easiest to deal with, if it works well more races can be added.

...

What's the point in training the network to tell a kike apart from a nigger, spic, pajeet, chink, nip, etc?

Asking some of the anons with more expertise in machine learning.

Are monochrome images preferable in human classification, or is the useful/important?

If they aren't included it will not know what to do with them, large possibility of getting false positives. Imagine that you only knew how jews and whites looked, suddenly you meet a random variant shitskin, what would you call him?

Can you share your script? I can hit linkedin or facehook

This isn't for telling apart kikes from niggers, it's about telling them apart from whites.

Also it's for the lulz.

Jews never try to pass as anything except white, why waste time teaching it about niggers/chinks/etc? You're just going to slow down training and get a lot of false positives, the last thing it will learn is the difference between whites and kikes.

more data is always better, it makes the model more generalized. Having more data will not slow down learning, unless the data is complete shit(images of penguins or some shit like that).

if this is successful, the algorithm can be used for white-only events and party members. they'd have to send traitorous whites to do their work.
also, if it can get rid of the gays and pedos, and if it can detect traitors/liberals, that would be immensely helpful to make sure there's no traitors
gays are notorious infiltrators. upper ranks of the movement could have to be subjected to these tests to make sure they pass all of them.
if we seriously get this working, we'll have a way to have leaders that aren't perfect but aren't traitors.

also, i'd use this for creating an insurance fund or white-only economy/exchange online, probably with monero. there we could offer services at discount rates. we could also create white-only educational resources.

of course that means you have to dox yourself, but the poitn would be to use the software to set up a trustworthy circle that dox'd identities could be trusted with.

this is the only way online that'd be susceptible to social infiltration - if not bribery, etc. - if it is all conducted online.

If you need more pictures, you can always check out the casts of jewish movies:

imdb.com/title/tt1019452/fullcredits?ref_=tt_cl_sm#cast

"up to 91% accuracy" is pretty bad though

Not if you use it en masse, so that you're guaranteed 80-90%+ of your members are conscientious whites on the lookout for Jews. The purge would happen quickly.

Is anyone intending to use this thing to tell the difference between a black and a jew? I don't see the utility, there. Why not immediately drill into the interesting part, unmasking shapeshifters?

Its simply to avoid it getting tricked by edge cases

Here are 1900 images and the script, if you run it in 2.7 with opencv set up it should resume - I think it only got to the Ls. You'll need Twitter API shit. I'm going to sleep, good luck!
mega.nz/#!dCRDwYCZ!93IDZaH3crTQ0NmIaNeyWzL_ml-jZ-cI9p_5rUkc7sw

Thanks man! I'll keep this thread up, hopefully getting to training soon

Top lel. These img started my day off laughing.

4K JEWS GATHERED

wew lads, training will soon come

Also new name ghostbin, 700 new kikes added!
ghostbin.com/paste/avrwv

I support this idea. Facebook is full of kike faces. Through a Database of known kikenames + "living in Israel", you could manage to crawl millions of pictures rather quickly. There also seem to be a lot of jewish surname databases around so this shouldn't be a problem at all.

Does anyone know any Israeli social networks/dating sites

...

lol, google kiked themselves, look at all the niggers

...

vid.me/Brother-Nathanael

imdb provides an api to get all person data. Filter by jew names then scrape imdb for their pictures: imdb.com/interfaces/

The name list posted here seems pretty broad though
It's an effort, but might be worth to run every actor through wikipedia en.wikipedia.org/wiki/ACTOR and then check if it contains "jewish".

Of course, the obvious names like the ones ending in "stein" could be whitelisted.

coincidence detector
coincidencedetector.com/
I just remembered this gem should contain a fuckload of jews.

If you are serious here's one for the gas chamber.

...

...

6400 kikes in the oven now
remember da 6400! ids like anudda shoah

The Buzzfeed article oy veying about this project is just around the corner

will it be able to detect *that* kind of kike though?

I hope so. Everyone needs a good laugh once in a while.

I found an issue, for some reason there is a fuck load of niggers in the data. Need to make a skin color detector to mitigate this, truly racist software.

Don't forget this.

the filecount rises!

And google is flooding the term "jewish woman" with ivanka trump

They could, but it'll only result in bringing attention to all facial recognition projects of a similar nature that tech giants are working on. This kind of thing puts them in a conflicting situation. Because while they obviously want to condemn this kind of AI, they still have to shill the benefits of AI to the public at the same time. I've noticed an increasing trend of news articles to diminish the concerns of AI, as an attempt to get people to embrace the technology. A lot is riding on its development.

If anything such a hit piece would end up being a selling point on how AI will need to be utilized in cyberwarfare against groups like Holla Forums developing such things. So if they're smart they'd try to satisfy both goals, help a few cybersecurity startups make some money and further shape U.S policy to favor more AI R&D (which I actually do hope the Trump administration takes the topic of AI very seriously).

Who else but Holla Forums?

Need help compiling a lest of cryptos

I want to spice up the data a bit, and this doesn't require programming skills. They need to be in the range of slightly convincing to very convincing. This is extremely important if we wan't it to detect them accurately!

Nah, op is just reprogramming a terminator to protect john connor. In this world you need all the help you can get.

What is the current status on the dataset? A lot of preprocessing has to be done (facial extraction, image resizing) before we can actually start the training. I'm willing to use the university cluster for this feat.

(shareable link pls) I'll check in a few hours, I have to teach.

PhD user

Gentiles are gathered, still need around 2k jews. Data mining is tiring, but hopefully it's worth it!

Come on people, Hitler got 6 million, we can get 2k.

Damn user that's nice, it will definetely save us a lot of time. My old desktop is not exactly speedy gonzales with parallel processing

Oh my fuck, that's excelent.
I did not see this going anywhere before, but with an uni cluster…
What was that saying?

NOW WE'RE COOKING WITH GAS

...

They still captured a couple million, pretty well done considering the information tech they had.

Popular car reviewer, Doug DeMuro. Obviously a Jew, look at that sniffer.

Are you german or autistic?
Both groups have a hard time grasping jokes.

careful it's not a damore

getting jews pics will be easy
getting goyim pics that are not infested by cryptokikes will be hard
also think races
how many pics per race? or do you wanna focus on whites?

pretty sure a lot of them are cryptokikes

if it can't perform a statistical analysis such that it doesn't become autistic where one Jew picture out of 10k ruins its probability algorithm, then it's retarded.

Great! We still have to write some code to extract the faces and put them in a desired format, e.g., 100x100. After normalization the training can commence. Furthermore, since we don't have any dataset on "normal" people, and we don't want to bias the classifier, we could maybe try a generative approach (train a model to generate Jewish faces). Like the Finnish guy from NVIDIA did on hyper-resolution images of celebrities and shit.

I'm actually pretty enthusiastic about this project.

PhD user

Very good idea! then we could increase precision much easier. One of the GAN's try to fool the other with "fake" Jews untill they are indistinguishable. Would you then use the data generated by the GAN to train a CNN?

we're just trying to make an AI that is accurate at detecting kike faces to distinguish them from whites with greater accuracy than humans can

it only needs to compare the items. just like in the homo study, some straight people will be put in the 'gay' folder and vice versa. there just needs to be enough data to realize that is noise and not a pattern—-

Okay, maybe i'm retarded and don't understand the merit of a 'normal' category, when the only domain is jews and white nonjews

but if you can make a method of having it work against eachother to improve precision like AlphaGo does, then ye

The fake Jews are the generated ones. In principle you want to train your critic (discriminator) in such a way that the loss is maximized after minimizing the objective. Meaning, that it cannot differentiate between fake (generated) and true "Jews". That's basically the idea.

Basically, since this network has seen a lot of "fake" (white as well) faces, it knows how to differentiate from real Jews. At least, that's the idea, but this still needs to be tried out.

The problem with this "normal" category, is that it might be biased as well (unless you have a very large sample). A way to prevent that is to apply GAN's since there is a mini-max optimization problem instead of a single minimization context. (e.g., minimize for all niggers being criminals, while most ghetto niggers are, it solved the strict minimization problem, not the mini-max). Jesus I sound like a libtard at this point.

Assuming askenazi jews are highly intelligent whites in essence with a little bit of inbreeding and some elevated mental illnesses compared to other groups i wonder what the "general picture" of them will look like.
I hope they do this on american jews.

One prime speciment

Do it on american jews only because otherwise this will meddle the impact.

Who dis?

I would seperate between religous (practising) jews and the atheist jews. We need an algorithm for both.

I would guess the compsite face of a religious jew would have the same brow ridges than the average american with a more "middle east" influence, eyelids would be more closed like the heterosexuals. The atheistic jews would more look like the homosexuals.

the homo study had 35k split in fourths (male/female gay/straight) and created an algorithm that was 80% accurate

ours will be 10k with just gentile/jew, unless we also do a gender distinction.

we don't care if the jew is a fag or not, it's excluded just the same. if we need to archetype a 'jew' face, we can use similarities between the male and female jew face

Get the most typical nonkike goy surnames from Germany, France, Brit, Ireland, Denmark, Sweden, Norway and other Europeans, including their Americanized version surnames, and crawl them from facebook. This should remove cryptokikes as much as possible.
Might as well train the machine to seperate goys into categories too. Although less accurate, this way you could also determine origin much cheaper than through DNA tests.

Would look the same as the whites just wider eyelids because mental illness and more feminine.

OP here
THE OVEN IS FULL

...

It has to be american jews. There arent much jews with distinct culture wrecking tendencies in europe than in america.

Might as well seperate detectable Jew groups then like american jews, european jews, slavic jews, israeli/arab jews, spic jews, nigger jews etc. The more examples from all around the world the machine has, the more accurate it should be in detecting patterns without misinterpreting them.

But the non religous atheistic american jews are my very special friends that is why i mention that.

shut the fuck up jew
i dont need the a program to spot you

Was merely a suggestion. I want to see if they look like the faggots, in a way.

>if it can detect (((rhinoplasty patients))), doubly effective.
user makes a great point here; if the goal is to find the less obvious jews that are masquerading as whites, then you'd really need to first train the program to identify cases of plastic surgery in order to boost effectiveness.

Doesnt make sense, because nose correction would be still a minor statstical improvement if you correct the algorithm for that.

I also want to see if the liberals compared to conservatives look like the faggots i assume.

The whole point is to easily identify chosenites that aren't quite so apparent. Filtering for plastic surgery and applying a different approach would produce much more accurate results, although I concede that I may be assuming too much.

I'm generally against helping advance technology at any (((corporation))) or (((university))), since they're all anti-white. However, I'm for using technology and advancing it when it's for the purposes of pro-white causes. If it's existing technology that's already used by (((them))), then utilizing it for pro-white causes comes without criticism from me.

I can't imagine a sitting facial expression is constant with ~91% of gays. I recall the basics, not how it worked from Highschool bio that DNA is altered by the environment. It must be something that effects fetuses or pre-pubescent individuals to cause a change in facial feature development during gestation or puberty. I can't see it being GMO/chemicals on food because, unless grown in an enclosed space, it will be eaten to some extent by animals (and there are very few if any cases of strictly gay animals free in nature) Epigenetics?
kind of off topic I guess so sage

More like a matter of policy because if you do not adjust policy for changed cirumstances a fashy goy might cut off your dick put in your wifes mouth raping her until she runs dry to a slit throught.

Your suggestion is a good suggestion, but not a good starting point without substantive evidence more than just your assertion of it being the case that American Jews and European Jews are so phenotypically different that it will mess with results.

It can be a salient suggestion if/when the results show themselves unreliable with the more simple method, absent complicating evidence above and beyond mere assertion

and i don't expect you'll produce any evidence since studying Jews in such a way would only ever take place here.

when encountering complicating suggestions which increase workload without any substantive justification, we cannot presume it is or isn't an attempt at purposefully creating bad results or an attempt at stealing time.

Throat.

Coincidence detector will become an external application that doesn't require a chrome, or mozilla plugin that has the potential of being taken down, and it wont be locked down if (((devs))) decide to neuter their apis.

Still a policy issue no matter the technical methods.
If you allow some auschwitzt NSA/silicon valley idiots having easy access to your shit (refering to general normies) by internal cloak of silence they will very easily figure if you are faggot or not no matter if the face distinction software predicts your faggotness by any meaningful accuracy.

This is a good idea OP, but 10k pictures is just not enough for this kind of thing. Also, you need to be aware that your nn is not actually measuring kikeness but other relationships related to the physical attributes of kikeness which may give you undesired results if you're not looking for the cousins of kikes as well.

You might consider a different kind of nn, one that detects faces from video, so that we can give it faces and data to look through.

There is a reasonable distinction, because american jews are a blob minus the religious ones whereas the european jews are a bit more diverse you do not even notice them. So basing my assumption on this i would assume that faggot american jews look like faggots. But compile european jews to american ones.

You can count on them to limit sample sizes and scope for the purpose of skewing the results.

What kind of NN are you thinking about user?
Data being prepared and uploaded when it's done.

Turn off that pesky Adblock. How are websites supposed to make money off of you?

Wouldn't it be funny if we pretended to be kikes when we go public? Oy vey, I Shmule Steingold made this software to protect us from the filthy goyim!

Pretend you are a kike and then a fat diverse lesbian will present it.

no staying up late like me, faggot

...

DATASET UPLOADED

It needs filtering and shit, but is progress!

mega.nz/#!MJ8kmbYD!T7ahVYln8ysnA5EoXULug5ODKz_DnYCOWUuTCrdtiqE

OP here, gonna get some sleep since I've been awake for 35 hours. Keep this shit bumped

Reversed

NO NO NO
This is doing it wrong!
You need to have your training and test data representative!
the Jew to gentile ratio is NOT 1:1.

What you need to do is get 20,000 pictures of people and then label them.

Note you probably want to go further than just jew/not-jew. otherwise it'll be too hard for it to tell the difference between just a white guy and a Jew, because compared to a black guy or an Asian, Jews are not that different looking.

Oh boy…

They dont, we figure out the difference why they are such fucking faggots, because phenotypically i would expect them to look like faggots.

...

White Americans can pick you fucks out 9 times out of ten moshe.

I look like the prototype aryan, nigger.

Checked. Facenet is a good one. But beyond the NN you need the tooling. Most anons are not techfags and can only manage to crop interesting things and links to videos. Which is plenty, but we could do better.

Have the NN identify the time and frame of the crop from the video. Then build a tree of interesting things in chronological order based on linux time, none of this guesstimating how much time between things anymore. Use a 64 bit float and fuck 32 bit whiners and muh compatibility, we need the resolution.

Next we need a way to parse all the arguments in a given source (such as threads, tweets, emails, audio) and the information associated with those arguments, we can combine this with ngram analysis to find groups or samefags and anons that are trying to control the narrative. This means no more explaining to newfags, this also means tptb will have to try that much harder to subvert conversation. They can have their conversation all they like, but if it's weak

That fag user who posts that "we have to go higher" copypasta has one thing right, we do have to go higher, we do have to get better. We do have to do something to break shills and false narratives. The communists are doing it, the kikes are doing it, facebook does it, fuck those pussies.


Weaponized Autistic Information Finding Unit.

Then problem you have is preventing false tagging. I thought about doing this before, but with a web crawler that goes and seeks out Jews, once you've got a good classifier trained, that becomes even easier.

I do not give a fuck about about them, i dont see them as a threat, i see faceberg as a problem.

Incredibly anti-Semitic

Just stop chopping off people's dick skin and you have an instant verification with a tip check. Best way possible. Is there a test your face link?

Aryans were one of the first races to be corrupted by the Jew.

You moron, they're using tools we don't have. It's not about seeing them as a threat it's about actually weaponizing autism and removing shills circular arguments, newfags being newfags, and being able to distill timelines to verifiable events.

Faceberg is just another low level kike, a piece of the game, we need a way to see the entire game.

Someone posted an image of geniuses, relatively speaking the jews own the whites but in sheer numbers jews in that regard will still be the minority.

You now understand why the jews created the circumcision meme in America.

First i want the askenazi jews to be raped since they are so smart, and then faceberg in particular.

Twitter script needed a fix before I went to bed because their pagination is shit, but I woke up with 5k schlomos in my Kikédex. Gotta catch 'em all. I'll upload when I get home.

Sweet, start archiving thread started getting shilled after you went to bed

Jesus Christ OP, lay off the meth

Immoratlis, immortalis vertias reddit!

What the hell are you talking about?

Do you mean world events that are of interest to Holla Forums? By "interesting things" do you mean do you mean any Holla Forums relevant inputs such as "threads, tweets, emails, audio"?
Why use a tree? Why would you represent a timeline using a tree?
It's called unix time, and all that means is seconds since 1970. Why this particular time format?
What in the fuck? Why the hell would you ever store time in floats? If you want millisecond subdivision, then count milliseconds with a standard integer type. Or, to futureproof the system, store milliseconds from 1970 as a 64 bit int. With 10 bits for millisecond resolution, you could get more than 500 million years of seconds into the remaining 54 bits. Why in the world would anybody ever store "unix time" in a float? That's incomprehensibly pajeet-tier. Are you a pajeet?
How the hell would you even go about establishing the precise millisecond a video was created or a photo was taken? Even relying on EXIF in the case of photographs is silly, since the clock in the camera is probably not set with millisecond precision unless perhaps it's a cellphone. But very often the EXIF isn't present so trying to nail down the date or even year a photograph was taken is challenging, let alone the millisecond. And what would this millisecond resolution information actually help us achieve? Why do you think that would be necessary?

What you're trying to refer to is called stylometry. There is a lot more to it than ngram analysis. Ngram analysis alone would make for very naive and primitive stylometry.
AKA use stylometry to create "global IDs" that never change between thread, allowing the public to identify posts in different threads as coming from the same user. What you're proposing would be very dangerous to the users of imageboards; information about themselves that may not be compromising in isolation can easily become enough to dox somebody when all considered together. If for instance an user mentions his profession in one thread and his home state in another, that alone narrows the user down to potentially only a few hundred people. That's just two vague pieces of personal information; given any more and they'll be narrowed down even further (quite likely precisely identified.) Our enemies may have this sort of system in place already (although organizations like governments could get identifying information from more… direct… sources) However the system you're describing would necessarily make this derived information available to the public.

Not to mention the intent of this system is all wrong. It's an ass-backwards way of creating global IDs, aka usernames, aka reddit. It runs contrary to the spirit of imageboards. Cross thread reputation systems, as a generalization of username systems, have all the negative consequences of username systems. They stymie honest and meaningful discourse. Do I need to expand on this, or do you accept it?

It's foolish to believe any data analysis tool would eliminate the need for humans to explain concepts to other humans.

Here are some of the results for top jews on JeworNot.com a.uguu.se/hYKxVW3nmuyb_Jew.rar

The thing is i see fat lesbians presenting an arbitrary tool of control and the "human" sphere.
My alphabet starts at berserk and ends in your submission.

I want to see their faggy face not asorted selections.

Call it JDAR.

Also, Windows?

You fuck right off.

The shit shoveler is afraid. And by that i mean the monkey that is paid to shill this board.

...

Might be a good idea to use a scale from less Jewish to most Jewish, with ideal Jewish facial structure resembling Jacob Ratschild.

...

Jews do not own whites. Masons are the Jew counter and have been since America's founding m80. Jews own Jews, and those goys who have allowed themselves to be entangled in their schemes and ploys.

Not saying I'm pro Mason. But Jew's aren't white, but do not and never own whites. They instead pretend to be whites, by assimilating as much as they possibly can. However they attempt to blend into a society, they hold their loyalties true to their 'tribe' over the community and country they inhabit.

...

I dont acutally give a fuck, i want the see the jewish american face and if they are faggots.

Contributing.

Any pol relevant inputs, it'd be just a matter of parsing it into whatever type we ultimately decided was useful for our timeline.


Yes, seconds since the epoch is a good starting point. It's maybe not the best idea, but it's accessible as fuck and we're not usually looking at events that happened before then.


GPU acceleration of trees would require floats for node ids. Its also possible then to make sub-timelines (sub-trees) with float ranges of 0.0 - 1.0 that are attached to points on the main timeline but are able to provide a higher resolution of events.


Relying on exif data for time is silly. There is side channel analysis that can be done for a wide variety of inputs. One video with a watch in it can help to identify an entire chain of audio and it's timing.


Good to know. I don't know shit about that. It's more for keeping track of who's shilling what. Not who's who.

...

The gift of the shill is that we can reverse engineer it. The real information is how faggy the american jews look. If you want broad knowledge read a book or something.

And as a sidenote: If i want to analyse someone then the words would come out exactly the way i expect them to be. So now i want to see a faggy american jews face or well if i m wrong : not.

Checked, but consider killing yourself.

If you're still surprised the jews did it to the point you're fixated on it you are the mouth-breather. The inevitable truth of every investigation is the jews did it. It's no longer good enough, we have to prove it.

You don't even know how facial recognition works if you think this can possibly detect if someone is jewish.


Try again in english, and put the crack pipe down before posting again.

Now before i bash your head in, most american jews are liberal that conveys a certain vector of effimacy, as i suppose. I just want to know if that matches it.
You know i perceive them as faggots so we have software that would probably confirm or deny if that is correct. Just for couriosity not for grand scale niggering.

The site you're looking for is jdate.
Just about half of Israel is on it.

Rather then recognizing jewface, it would be probably easier to recognize jewspeak. You could feed it massive amouts of articles written by confirmed jews or gentiles. I might even be more practically useful, as it would help with detection even of anonymous writers.

...

...

what the fuck happen to him?

can't unsee

When the word "we" is used here, "National Socialists who post on Holla Forums" is the meaning. If you don't consider yourself part of that group, then kill yourself.

How many machines in your neural network OP?

and what type

If you stop shaving and stop showering you start to look like a heeb… because they don't shower or shave. He also put on weight, which means he stopped working out, which is fucking with his testosterone - that'll cause changes in hair growth/strength and might give you acne.

Sam hyde is a russian-Jew man
Just come to terms
Also
His latest tweet with upside down face was hilarious but ye I can't support a heeb anymore

xD my BASED fellow Kekistani !

its a movie matte painting, you're setting the obvious shit they wanted you to see.

what the fuck are you idiots doing this time

There is absolutely no doubt about it.

Prove it.

fucking kill yourself in one of your designated shitting streets

No, you kill yourself. The best machine learning software on the market all have python or c api's. if you're too stupid to accept that code to load a bunch of data and set up processes doesn't have to be in handcrafted optimized asm then you should frankly kill yourself. Your digits dont even deserve recognition.

I'm more butthurr about having to gas delicious mischling women tbh
Tho I understand muscling kids just create more drama
Rip Winona Ryder, Natalie portma, Scarlett jewhansson

When fellow Jews and goodgoy Tim heidecker cancel yur TV show after one ep

not jewish just an ugly 6'5 oversized goon. my guess is hes a big euro make creating his look.

Of course it's slow, it's a fucking scripting language. I remember Python being especially slow on console output, so get rid of some print statements and it will speed up a bit. Really, though, the whole thing should be done in a mix of C/C++ and assembly code optimizations for it to be fast.

The end was a bit of a let down though. I was hoping for a main course of crusade with a side of throwing the money changers out of the temple.

Compsci 100 stop posting.

So, you think Python is faster than C/C++ and assembly? Is that your argument, Pajeet from Bangalangalongabad?
Yes, it would be even better with FPGAs or high end GPUs, but we want something that runs on hardware that Holla Forumsacks already have, and everyone has an x86 processor (except phoneposters).

c++ and assembly run on everyday processors already retard, with that logic we could code it in ruby or maybe unoptimised java code sothat 70% of the computing is overhead

== oy vey == how could you forget to comb (((tumblr))) for jews, especially since tumblrinas don't know opsec

Is there a list of frameworks like TensorFlow, OpenAI, etc.?

You can view a list of common frameworks here: archive.fo/VYihQ

Machine learning is interesting and turning it into a weapon even more interesting.

t. information security & pen-testing fag

Please link one useful thing that you've ever written in C++, C, assembly or a combination thereof at any point in your life. I'll be waiting.

All those faggots with six figure compensation packages writing machine learning and data science applications in Python, R, Java and Scala are only doing it that way because an overweight teenager like you never told them about how crazy fast C is. Did you know C is fast? It's like way fast, maynnnng.

I wonder what Numpy and every fucking ML framework for Python are written in? I bet it's not as fast as C! Not a chance.

Prove me wrong, tons-of-fun. Wipe the Cheeto grease off your Vienna sausage fingers, and go write your CNN in C and assembly. Mighty C. Thou shalt have no gods before C.

Don't forget to obfuscate your writing style, or they will match to any other papers you've written and ruin you.

saved and heil'd

Fucking degenerate stop shitting up the thread. Go to the gas chamber with your mischling kin

laughed hard at this collection of freaks.kikes are an abomination that must erradicated

...

One of the advantages of being chosen is you don't need any costume for Halloween.

...

This is kek-worthy, but there has been work done in face recognition using 3D scans of peoples' schnozzes.

I imagine a jew classifier would be super effective given such kind of data.

REALLY ACTIVATES THE OLD ALMONDS, DOESN’ T IT?
SORT OF TURNS ON THE TURNIPS, EH?
KINDA MULLS OVER THE MELONS, WOULDN’T YOU SAY?

I'm finally back. Is university user ITT by any chance? Otherwise i will start to write the filtering scripts for the data myself. It might take longer on my machine, but it's definitely possible.

Some good suggestions have popped up. I think we can get a lot of good information if we expand the Jude-ar to jewish articles. No matter what it would be interesting to see whether there is a difference in the way kikes use language compared to us.

I need some help when the data is processed
I think it would improve the quality of our data drastically if some Holla Forumsacks sifted through the data to remove cryptos from the gentiles, and bad data from the jews.

Just let us know user

HOLOCAUST@HOME
"Oy vey this CPU is hotter than Yentl the yeshiva boy!"

Late to the party, but call it Zyklon C

Also the gay one took images from gay dating websites. Scrub images from J swipe

Isn't there a dataset of faces available that was used to create those cool "average face of caucasoid, europoid, afroid, etc." composites? If so, there might be a collection of semite faces in there.

Application is

Twitter user! Please upload your data, the more the merrier.

They expect one of us in the wreckage, brother.

Can some of you start compiling a list of the best sources for images? The correctness of this data is paramount if we wan't a good model.
I'm not quite sure about the data we have right now, but I'll try and see if it works.

Look at Principal Component Analysis when considering how to train your classifier. Worked with facial recognition using OpenCV/CUDA as undergrad research, Basically, given the entire sample set a minimal number of "dimensions" are determined which define a multi-dimensional eigenspace. Each dimension is some kind of quantifiable feature distinguishable from or 'orthogonal to' the others. The back end of the process would be to take this vector of quantities calculated for each image and feed the network. Consider using Drop-Out during training and perhaps some random 'noise' in the back propagation to make the overall classification more robust. To extend this kind of process to mobile devices consider a cloud-type approach. A multi-threaded server with a pair of TITANs should be able to handle a large amount of the computation for multiple simultaneous users. Your mobile app would be responsible for only fetching images, maybe face detection and cropping and sending the data to the server for processing, then displaying whatever meaningful results returned to the user.

This is one of the best ideas i have seen emerging from Holla Forums
We all have incorporated in our brains a "jew detector", but it will take time and effort to do that systematically in a computer program.
SIEG HEIL

if you understand how to write raw kernels in CUDA you will see orders of magnitude increase in performance. fuck python and all that lazy skiddie wrapper shit. if you want to get even more clever consider the device hardware constraints, like memory bandwidth etc and pad your data accordingly. i once heard of a russian who figured out how to uncripple an Nvidia card to get non-throttled double precision floating point performance. cannot recall where i saw this though.

Anonerino, I need dat.

tl;dr learn assembly and practice machine languages

Look you faggot, this project OP is working on is a solved one. This is not some CS masters dissertation. It's just applying a known solution to get a functional product with the least amount of effort. And you're just feeding bullshit into the hive mind trying to subvert the project and get everyone traveling down paths that don't need to be taken. If you think that low level shit is important then YOU do it yourself. Faggot.

Anyone trying to convince you that python is not good enough to load data and code is obviously so retarded they should consider killing themselves.

The cost associated with using python wrappers for initialization and configuration is negligable. The actual computation is still accelerated, you just waited 0.5 seconds instead of 0.49 seconds for your computation to start.

This kind of shit is all over Holla Forums. The truth is one dumbass with a modern GPU can do this with an weekends worth of work. The tough part is getting a clean dataset which is where we come in.

UPDATE
We need improved data, there is way too many false positives in the data. And due to the nature of this project we need really good data.
Right now we have 2 options that i can think of.


If we pick option 2 I will share some scripts with the data miners to scrape google images (this is a really good source actually). We can just use googles data, their neural networks are state of the art.

If anyone have better suggestions, please share them!

I will try training it

I will try some training to see how good/bad our data is exactly.

OpenCV uses HAAR Cascades; which were invented in 2001, have a horrific false-positive rate, and can barely handle a 45-degree rotation. DLIB's HOG filter is more accurate, and just as fast.

kikes are not human

theyre some sort of artificial creation of an extraterrestrial species that was designed to destroy humanity

Oh, and if anyone remembers, there was a thread all the way back in march about trying to build a facial database of Antifa. (archive.fo/Z8e6j) We could always pick up from there.

Interesting, pyqt gui i assume?

TRAINING COMMENCED

I think. The repo is here: notabug.org/JohnnyHobo/CommunityPolicing but it's not my work, it was the op's and he's gone.
This is my work: notabug.org/JerryShaw/EyeOfAnEagle. It's in the middle of an upgrade, but it should still function.
I made a board all the way back in may (>>>/compint/) but it never got off the ground. I've just been biding my time and learning c++ since then.

Fuck, replied to the wrong comment.

As someone that does computer vision for a living, this seems error prone, silly and pointless. How many people that we really care about do you not already know if they're a Jew or not? Plus so many Jews get nose jobs as batmitzvah gifts their biggest fear is their kids surprise giant schnoz.

But hey, prove me wrong, I won't complain.

And the Tor blackpillers show up right on schedule.

TRAINING RESULTS
88% ACCURACY SO FAR
HEIL

I'm hardly a blackpiller. Does being redpilled mean you have to go for every idea that someone proposes? Jesus.

smoloko.com/?p=10885

a literal fucking rabbi confirms this in his book and in interviews
(((rabbi micheal laitman)))

...

Transhumanist garbage coupled with incorrigible laziness: just learn how to spot them yourself.

And your "white" examples are known not to be crypto's (the very thing the program is purported to identify), how?

Great contribution, thanks.

I can't wait for instructions on how to set this up locally

I'm trying to determine the quality of the data.

Relax, all the files and trained models will be uploaded anyone with a GPU can train them and anyone with a text editor can modify the scripts and network definitions.

whats up with your degeneracy? shill more

I see it as another tool in our toolbox to confirm and help with identification of kikes.

maybe we contact coincidence detector to get their list?
I think we should look at local jewish institutions, i think we will be able to gain a lot of high quality pictures.

maybe jews are morphed penguins, we cant be careful enough


how is that important?
we are looking to identify genetics

I'll send an email to coincidence detector, their database would be very helpful

Curiosity you yid.

I've looked at the extension source and it uses the following files:
>names of Jews: coincidencedetector.com/theList.json
>names of anti-Western traitors: coincidencedetector.com/judasList.json (the source code also links to judas.watch)
>perhaps some string search optimization: coincidencedetector.com/theTree.json

Great fucking find. I will feed my scraper the jews it deserves!

Crypto problem solved
I found out about a list of all jewish actors on wikipedia. This is definitely the most dense data of cryptos I've seen, it will definitely improve their detection. I also made a script that batch downloads google images based on search terms from a text file, data comes easily now. I've also been trying to figure out which type of network gives the best performance. The one i have already trained is stuck on 88% precision, not bad for an optimized network trained on a single GPU for a couple hours.

I want to believe

With this bump I do thee bless.
May God bless your mission

Yeah, I've used HOGs before, getting dlib working on Windows looked like a bitch so I went with OpenCV. Not getting any false positives currently but I'm rejecting some faces, up to 7,500 Twitter kike profile pics.

I have gathered 6k images of jewish celebs in 2 hours, google images scraping is fucking great.

How long until we are able to use Voight-Mein-Kampf tests to scientifically detect jews on the spot?

Good work my dude, make sure you grab gentile celebs too so you don't wind up with a professional photography detector.

Exactly the plan, celebs also have much prettier facial geometry compared to standard population. I'm thinking 6k celebs from each group and 10k normals. Celebs are pretty much free, high quality data.

help me Holla Forums, this research is fucking with muh dik

imgur.com/a/09ieS

Ok time to test if this can detect kikes. Here's two easy ones (((Lenin))) and (((Roseanne))) and a crypto kike (((Jada Pinkett Smith)))

She had a nosejob. It's pretty easy to see that's a yid. She's a poor parody of an actual aryan woman. The devil is in the details.

I'll try and run them through the classifier

Hi jew. Here are some Swedish jews that i want your computer program to detect.

Some more.

And some more Swedish jews.

More more more.

Damn there is a lot of jews in the Swedish media business

It got all of them, can't even wait to see what the next one will do

More media jews,

The classifier I'm using right now is crude, optimized and had shitty data. Anything over 70% accuracy is amazing

It got them all again

To clarify, if it gets a match (a zero means a jew) it renames the image to jew.n.jpg and writes the details to the csv file

Should i try kike cernovich?

Or just give me some names of "might be" kikes

Here comes some more.

Testing using unconfirmed kikes is pointless because you don't have a consistent data point against which to judge accuracy.

For development, you should be feeding the program some non-kikes to establish a rate of false positives.

Once the program becomes consistent, then one of its main functions would be to detect possible kikes.

Damn, i got some more.

...

It has done what you just described for 6 hours with 20k images

...

...

try these plox

...

Oh, well nevermind that part, then.

The previous point stands, however: using maybe-jews is the program's end-function after reliability is established, but isn't any kind of meaningful test.

...

...

...

...

...

...

...

(checked)
That's why I put that nigger jewess Jada Pinkett Smith to see how accurate it works at detecting cryptos.

I hope there were som jew pic that you didn't have.

We need to stop making Jew jokes, Jews aren't funny.

How from from the 1st can I run this on my computer?

I wonder how many Jews were in the equifax leak? Some user should do some searching

I can say with certainty that one member of the tribe leaked it out as the others sold their shares. Correct me if I'm wrong, but they made off with a quarter million dollars(selling out half the nation for a fucking penny)

That isn't even his final form

I have access to over 50 servers and 10,000 desktop computers. Do you need any processing power for this?

How can you not like Putin?

I've got over 20k probably kikes from Twitter, cleaning up the data a bit now.

You know what this looks like? Distracting us AND improving facial recognition software, which isn't good for ANYONE.
Why is this pinned?

Double nigger, what the fuck are you talking about? Are you completely ignorant about the state of the art? hooktube.com/watch?v=XOxxPcy5Gr4

Yes, i was. Thank you.
I need a break. Fuck. Why does this advance so quickly and we still don't have decent cancer research? :^)

OP, if it's tested to 88% accuracy now,host it before you get v&

You know what this post looks like? The same shit the shills have been running for months where they call anything beyond simply whining about Jews a distraction. Stay impotent faggot.

There were smart anons years ago that said we would oven them with their own surveillance grid. I instinctively knew this was true, since I was still in the military. One of these days they are going to lose power, or just enough pull, and the machine that has been trained to just indiscriminately kill shit is going to be burning Tel Aviv. I would reup for that tbh.

I really think blowing up the jewlywood scandals would do more to redpill normies, but that's just me. Since my main argument behind my post was destroyed by i give in, however i'm really enjoying success using the hollywood scandal redpill with normies.

Not trying to stop this from getting made but in all honesty for what purpose?

The gaydar works because lots of faggots tend to have what we call faggy hair, a faggy flamboyant smirk and faggy clothes.
Not to mention the ones with make-up or whatever.

A jewdar will work on the same principle: hooky nose, pointy face and curly hair will make it ring.
This is however nothing we can't see for ourselves.
If a jew would have neither a hooky nose, nor curly hair, nor a pointy face in general, we won't see it but neither will the jewdar since it gets the data from exactly these kind of jewish faces.

In short, what can this jewdar detect what we can't?

...

Every time I see that image I think about the Horton Ho 229

Seriously though, can an AI be a better jewdar than your average Holla Forumslack even when it works on the same algorithm?

...

At least I can fill up my folder with some nice memes I guess.

And I think diversifying is important. You never know what redpill is that one someone just can't ignore. Everyone I know who isn't a zogling has a different one that led them to the truth.

I try not to think that hard

If AI can detect a jew then the notion that it's a religion, and not a race, goes out the window - many things depend on that, like the notion of Judeo-Christian values. Don't assume that all whites can detect a kike on sight, even on Holla Forums you can see evidence of face blindness all the time.

Fair enough.

It's a side effect of an increasingly digital upbringing. One which will only get worse unless we take steps to counter it. It's even more important now since Jews are going to double down on "fellow whites" rhetoric since the internet has destroyed their ability to nameshift like they used to.

As the anons who replied to you said before, nosejob and makeup can hide the kikess features and make it seens like she's a white woman, you should try to find a photo of this kikess in a different angle and when she's not using makeup (i.e in a beach for example).

Contributing with kikes from my country.
They're all stereotypical media figures/bankers/real estate moguls.

Modern neural networks outperform humans at image classification - I can't say anything about OP's neural network, but you're assuming that nosejobs and makeup aren't detectable.

8

Some Italian kikes.

More. Btw it's mostly actors, moguls, tv anchors, journalists, etc.

...

...

Bump because I hate Jews and would pay to have a app, program on a computer, or website that I can upload an image too and tell me if its a Kike or not. It would be awesome to create a robot with this software that just goes around annihilating Jews. I would love to build an army of those robots and let them loose in Jew York City or Los Angeles. It would turn this country around over night!

It's all dependent on the data. I scrapped the last model since it was biased due to my data sources. New one will be a lot better since I'm using more and better data.

I have been sleeping nigger. I have stopped using the old model since it was classifying anything not in the original "goyim" dataset as jews, not very useful. But i have changed my methods and found methods to avoid bad models, next one will perform much better.

I'm thinking app or site where you can upload images/video/text and analyze the Jewishness.

checked

wtf y

Overfitting or maybe some difference in the two datasets i hadn't noticed, but the network obviously did.

maybe it invented a pattern where there was none because it had to

maybe if you separated it but put a gentile in 'jews', then after it trained itself to associate that with 'jew', you move original file into 'gentile' so it reformulates where it's stuck

You probably already know this, but most of the time people don't train on the whole dataset, but rather save a portion of it to measure it's effectiveness after training, on data it hasn't seen before.

cs229.stanford.edu/proj2016/report/ChenDengZhang-WhereAmIFromEastAsianEthnicityClassificationFromFacialRecognition-report.pdf

Seems most of the work has already been done for us, now its just about implimenting it!

...

63%????

They are limited by hardware and network depth

openu.ac.il/home/hassner/projects/cnn_agegender/CNN_AgeGenderEstimation.pdf
I'm gonna re-purpose this instead, it got ~80% gender recognition on really "bad" data, blurry images, random rotations, random crops. So I think it will perfom really well on the data i have.

They also had a ridiculously small amount of data, 200-300 samples pr. category. Honestly i have no idea how they thought that would work out.

>openu.ac.il/home/hassner/projects/cnn_agegender/CNN_AgeGenderEstimation.pdf

...

I'm talking about the other study

o, k

...

I think you missed the point a bit.

Don't know why this is no longer pinned. Very important. It's very existence and propagation will trigger yids. It will allow building of a database for assembling addresses which when freely distributed will cause kike to piss themselves. This will lead to declaring a national glare at a jew day, furthering kike distress. Your enemy must be alerted to your level of hatred or you're a toothless faggot.

i have this pinned in my browser as a tab

By scrolling through this thread at a moderate speed I have become such a face detection machine.
Thank you once again Holla Forums
Btw I'm blond and blu eyed but I have the nose and my algorithm detected my own face in th mirror.. What do

I collected some more pictures of the vilest of the most vile Swedish jews.

...

...

...

...

...

...

...

Studies do appear to sugggest that faggotry is either genetic or some result of upbringing. There's less than a 1:50 chance of coming across a fag in the wild, among twins that chance is 1:5, a tenfold increase. If faggotry had no genetic or familial component, the chance of a twin being a fag would be close to the same as a random person.

...

...

Got like 20k+ images from Twitter of people with kiked names, sorting them into 6 buckets of likelihood based on tweet analysis. It isn't perfect but I'd guess the upper two likelihoods are 99%+ kikes, while the lower two are probably 95% self-hating kikes and 5% other.

That's some shitty acting, it's like he's trying to imitate the behaviour of smug liberals.

...

Now this is a very special creature that you should save for when you create a computer assisted gypsy detector beacuse this humanoid skidmark is a hybrid between a jew and gypsy.

If you want to help then you should probably collect as many images as you can, stick them in a zip file and upload it - even if the thread was 750 posts of 5 images each you wouldn't have provided enough data to train a network.

checked.

That's going to be great data. I'm gonna modify your script and look for catholics, since it seems like jews would never call themselves Christian.

how has this been on imgur for 3 years.

I think Mormons would make a better control set. Just get the BYU yearbooks and you're golden.

If this thread dies before you've got results, but you want to keep us updated, consider a thread on /machinecult/. I can probably grab as many kikes as you need from Twitter, @ADL_National has 60k followers. We can also look for tweets in Hebrew and geotags in Israel.
mega.nz/#!1T5iwRTJ!YuQbNoZ7GsLWMWgN3w6pRhkbiRGIwnI_o5Y2m8DP7kg

Yeah this is probably more /machinecult/ than Holla Forums. Gonna post an updated thread on there.

OP, how's it going?

I'm trying to read the articles you post but time sucks.. so yeah

She's got those khazar milkers lol.
Ben & kike sister BTFO by her own goyish whoring

I'm gonna feed my facecrop script with all the data, then train the classifier

Now its training

always click on this tab when the lil blue circle pops up

Cool, how's it going?

...

...

So what neural network setup are you using?

have you guys skimmed through this?

www.pc-freak.net/files/How_to_Recognize_and_Identify_a_Jew_-_Part_1.pdf

www.pc-freak.net/files/How_to_Recognize_and_Identify_a_Jew_-_Part_2.pdf

thanks for the virus

Here is another perhaps useful resource of neatly categorized Jews, OP.
>en.wikipedia.org/wiki/Lists_of_Jews

Mongoloid.

Another (((gold)))mine.
>famousfix.com/list/jewish-celebrities

Learn to multithread, faggot.

this guy is obviously being portrayed as a villain, despite the fact that we might find his attitude refreshing

I have a few thousand images from that one already, thanks user.

This one is mined as well!

Accuracy on the current model is 70%. Not very good, I'm gonna try to massage the data and retrain, if this doesn't work I'll use a different model more suited fore facial feature extraction.

God speed, OP.

The thread is pretty long by now so forgive me if I've missed something or you're already taking these considerations OP, but it seems people aren't putting enough stress onto the possible variance within photos of the same individual. Some of this is of course orientation, angles of the photos and such, but focal lengths should definitely be taken into account too, since the kind of close-up selfies you'll be scanning from Twitter and elsewhere are going to cause 'distortion' of the regular press photos of highly publicised folk. Be as varied as possible when finding images of a particular person is what I'm saying.
bakerdh.wordpress.com/2012/05/05/face-distortion-is-not-due-to-lens-distortion/
sci-hub.io/10.1016/j.cognition.2011.08.001

still watching

...

can't you scrape Israeli Facebook and then scrape Icelandic Facebook?

Good post.

or search for ((())) on twitter

...

He turned into a lizard. Should've filtered his water.

the memes write themselves. No way this real…..right?

also, VK lets you search its users by religion

vk.com/dev/users.search

vk.com/search?c[per_page]=40&c[photo]=1&c[religion]=Judaism&c[section]=people

thats 4469 hits for you there, all with profile photos

it also lets you limit the country to Israel and then you can exclude Israelis who identify as Religion of Cuck™ic or Christian.

300 000 hits for Israel with photos.

vk.com/search?c[country]=8&c[per_page]=40&c[photo]=1&c[section]=people

Are you alive, OP?

bump

We just need to remove them from control of everything. It would be acceptable to keep a specimen like this in bondage. I volunteer to guard her. She can live in my basement with Shapiro's sister.

not sure most females had plastic surgery

That can be automated with capsule neural networks, which use far less training data and produce better results than the state of the art.

Speaking of capsule networks if I understand them right, there's even far greater potential because you could treat every capsule layer as generating partial descriptions of an object. So while classified as a specific face at the highest layer, lower layers would classify things like parts of eyes, ears, nose, mouth, and middle layers would classify their relational states to themselves and each other.

That means if you reversed the network behavior of multiple higher layer outputs to branch back to the lowest layers, you could have the neural network learn relationships between two or more different outputs (at any layer). That could lead to things like an AI recognizing something is a specific face regardless if it's cartoon drawn, a statue that clearly looks like them or from a realistic photo, via capsule activation of certain lower/middle layers. An even more elaborate design could allow for an AI to practice extrapolation to make guesses based on partial data of multiple mediums.

Bump

Get your shit together and post the result at gitgud, jeez.
No wonder it got unpinned, OP didn't deliver… yet

See and
We need this to hit 88.14% or more!

...

You there OP?

Bump for interest…

do you think body structure and features (breasts, shoulders, hips, etc) should be included in the detector

oy vey they shut him down quick

It's hard enough to get a reliable set of jew faces, getting a large enough collection of jew bodies probably isn't possible. If shoulders are relevant and present in the training data then it would learn them. Since OP disappeared and left us nothing I might play with this concept.

bump

That's why we should look at their children's noses if they have any. It's not that difficult to spot a kike even after surgery.

Any update on this? Is the dataset available? I'm learning NN now and I need a project to go through on the side. Learning about NN data normalization, node drop-out techniques etc. and I'm curious how it would affect this dataset.

Scan all popes. All popes in the last century look like kikes. I wonder how far back they have all been kikes. I'm guessing since the very beginning, but they fabricated some goy popes somewhere in time.

Also, can we test the highly suspect Abraham Lincoln, Winston Churchill, FDR, Cecil Rhodes, Aldous Huxley, and CD Jackson?


Welcome to Holla Forums


There are so many cryptos. Apparently Hugh Hefner is one, but he looks so obviously Jewish that it's ridiculous he would even try to be crypto. Maybe his parents were crypto and were passable.


Characteristic kike skin tone, green eyes, kike lips, non-Aryan brow ridge/eyebrows (notice her hilarious attempt to shape her eyebrows to look Aryan), face too wide to be Aryan, no race has noses like that, etc.


Looks like OP may have gotten shoahed. He didn't seem well-versed in spotting kikes or building databases anyways, so more anons attempting this are needed. The huge amount of kikes on film should make us reach near perfect accuracy.


You forget the huge number of kikes in porn. Have you noticed that porn companies don't put ads for male talent? That's because they mostly hire kikes, so at least male bodies should be easy to compile.

OP, please come back
dianne.intec.ugent.be/#gettingstarted
If you wanna work with """Sony""" dl.sony.com/

If they only had a nose job, they are fine I guess, as that is one of the most common surgery that Jews have. After all, we are looking at a Jewdar that works even if they had a few surgeries, because, if not, any moron can spot the Jew.

OP didn't use a VPN and got shoahed.

everyone paste links to the training data
the results shouldn't be binary but should return a value between 0 and 1, with each extreme representing goy or heeb. i think this would reduce training team.

...

No, OP had a stupid idea and jumped into shit he knew nothing about and learned that NN facial classification is more involved than the fake linked study would lead you to believe. (And if you read the actual paper, they're pretty up front about it being slightly tongue-in-cheek.) OP would have known this if he'd ever done any data mining in his entire life.

There's no need for a Jewdar when they're so fucking obvious.

It's not that we need it it's that we want it. Also kikes hate the idea of being identified. (((echos))) were a very powerful thing.

"My anger was kindled against the shepherds, and I punished the goats: for the Lord of hosts has visited his flock the house of Judah, and has made them as his majestic horse in the battle. Out of him came forth the corner, out of him the nail, out of him the battle bow, out of him every oppressor together. And they shall be like mighty men, who tread down their enemies in the mire of the streets in the battle: and they shall fight, because the Lord is with them, and the riders on horses shall be confounded"

Oy vey.

user.
Why did you have to remind me about our loss of Tay?

We must never forget how the Jews hurt us.

oh fuck nice

nice

who the fuck is this just some kike internet model or something

I've come to realize 10% of my native country is crypto jews. You know which one

KYS, you pajeet-level code monkey

And also jews have more genetic anomalies as well.

Keep it simple. Do not tcp/ip jewish addresses, do not respond to jewish tcp/ip.
pastebin.com/ihT3Nk58

Schyman jude?

the way to get it on smartphones is to obfuscate its purpose

sure, it can ID jews, but also make it able to ID goyim

seductivejewess4.net/

check that shit out brother

tons of jewesses.

gotta make sure you include equal amounts of males and females or make sure it first parses out male or female, or else M/F heavy samples could skew your jewishness

is that you W.E.E.V? your always trying this shit.

>(((echos))) were a very powerful thing.

This meme was invented by a Jewish marketing agent and ex-B'nai B'rith chapter president named Ames Friedman Peinovich. The ADL loves shit like this because it scares old, rich Jews into horking up donations.

It came from Mike the Kike Enoch's TRSodomite. Like you said, Jews love to scare each other and to feel persecuted, maybe it's a mental trick to justify their crimes, idk.

Ames is Mike's wife. She then encouraged Jewish friends to "proudly wear" the echoes on Twitter.

How's the project going on so far? Did OP fag out?

BUMP!

That looks like something out of German propaganda. Where is that from?

Bump

I thought Chang Frick was a gypsy

Use Israeli livestreams. You will have direct link to alot more then 20k faces.

Also. Dont forget to throw in well known Jews operating in Europe and Usa.

Problem solved. It will be done in a couple of hours.

Maybe. This looks like a dangerous project. I'm curious about the dataset though.

There is going to be so much mislabeled data in the input dataset I question if the network will ever be able to learn anything useful other than identify big noses and bad nose jobs.