[ / / / / / / / / / ] [ dir / cute / fur / htg / lovelive / misr / wai / waifuist / wooo ]

/pol/ - Politically Incorrect

Politics, news, happenings, current events


Comment *
* = required field[▶ Show post options & limits]
Confused? See the FAQ.
(replaces files and can be used instead)
Password (For file and post deletion.)

Allowed file types:jpg, jpeg, gif, png, webm, mp4
Max filesize is 12 MB.
Max image dimensions are 10000 x 10000.
You may upload 5 per post.

Dear Fat Kimmy, there's a French Banker married to a grandma who's plotting to replace you with Yeong Ten

File: e8f3d62d902f5b5⋯.png (223.47 KB, 1366x735, 1366:735, SOON.png)

887024 No.9843629

We've all talked about it, we all want it. Nobody wants to be the one to make it, so let's do what we always do: Decentralize.

The idea is simple, everyone has a database of the information they've gathered. They host said database on their own machines that they can run their facial recognition queries against; if they want to share they can use Tor or just copy-pasta the data folders.

Program Requirements:


>Reverse image search and facial recog software

>Social media search

>Web browser and scraper for more hands-on searches

>one click web archive

>easy to use GUI's and Wizards

>Open Source

>tor ready

>easily expandable, modifiable, hackable, etc. (written in QT, Python, BASH)


Any tips, suggestions, thoughts, etc. please clapshare ITT.

I'm currently working on the GUI the part I've always found the most difficult and it's coming along very well. I did all of this last night and today, the GUI is about 50% finished; backend should be rather simple.

887024 No.9843642


887024 No.9843713

File: b514f0cd4072ee7⋯.jpg (397.21 KB, 1241x768, 1241:768, more-collusion.jpg)


I do recall The List meme tweeted as a no-context method for scaring the piss out of kike / traitor journos and politicians a while back. Was bretty funny.

I like this better. Have a bump

887024 No.9843729


Well there has been some plans in the past to create an onion site or distributed chan in the past. I'd go more for the distributed chan with /pol/ on it then just advertising your folders on WN, because it creates plausible deniability by there being a /v/ section or /b/ section. Only problem then is insert CP and its now on your PC.

So you'd need a crypto chain to ensure messages keep their contents which isn't hard and or could only be selectively read so you mitigate the CP directed to your IP. Just like time versus space you also have privacy versus coverage. I would say work on the actions UML before making a GUI however though your GUI seems only for computer vision and not hosting.

I'll be looking deeply into blockchains for automation a few weeks from now. Are you using OpenCV for your facial recognition?

887024 No.9843749

bump for sticky

887024 No.9843865


Nntpchan already exists anon. I2p and onion based chans are out there if you look. They just need skilled fags to come to them

887024 No.9843891


>The List is a meme

Have a sample.

Adam Kinzinger

IL Congressman, #NeverTrump

25566 S Keating Blvd

Channahon, IL 60410


Karl Rove

Former WH Chief of Staff

1408 Circle Ridge Dr

West Lake Hills, TX 78746


1333 New Hampshire Ave NW Ste 600

Washington, DC 20036-1532


Deborah Wasserman-Schultz

Former Chair Of DNC

4479 Fox Glove Ln

Weston, FL 33331

(954) 384-2225

Jerrold Nadler

Democratic Congressman/Attorney

315 West 70 Street 3C

New York, NY 10023


Adriano Espaillat

Democratic Nominee for Congress

62 Park Terrace West A87

New York, NY 10034


Daniel Vila

Green Party Congressional Candidate

192 Claremont Avenue

New York, NY 10027


887024 No.9843895


It's not a meme. You think those voting machines just un-rigged themselves?

887024 No.9844024


>Are you using OpenCV for your facial recognition?

That's the plan.

As far as the servers go, they really shouldn't be talking to each other too much at least not at first, the main thing I was looking for was pulling from other databases rather than pushing to them; word of mouth rather than "Here's a list of repos, download away". Some people will still get hit, but it is much safer than allowing unfilterd inputs to a database. Nudity filters also work, that might be something to implement down the line with a whitelist and hashing, but people can still hide pornography in binary code so anything short of setting up some kind of website and using a webscraper on top of that as well as screen-shotting the images themselves AND applying the nudity filter + whitelist will still fall short in some areas. I suppose a community PGP voting system via keysignatures would also help to filter out the smut.

Important thing is to get the client's created and a database structure setup so people can start working with it and getting used to it while the networking gets sorted out.

I also need to write the (((Terms Of Service))) so I don't get shoa'd. Don't do anything illegal with it goyims.

887024 No.9844074


I do find the blockchain idea interesting though; but we're talking videos, documents, web archives, and pictures -not just a bunch of hashes in a ledger.

Maybe a blockchain could be used to hold PGP voting info, names, and other information.

Actually, discarding the images you could very easily do a distributed dox database with a blockchain. Since a dox is generally just name, address, phone, links to archives, and a few comments on their activities it shouldn't be more than a few hundred bytes. Fuggging great idea with the blockchain anon

887024 No.9844219


name coin?

887024 No.9844276

Nigger use something based on block chain technology then checkpoint on the bitcoin block chain to ensure data integrity. Don't be a pleb

887024 No.9844279


Monax, formally Erisdb look it up

887024 No.9844416


Add a way to track organizations and a persons affiliations with brief explanations of each affiliation. Would make it much easier for people to connect the dots, especially if you made it so an affiliation has to be another person or organization in the database. This way everything is easy to verify and understand, and it could prevent duplicates.

887024 No.9844585


I've made ruby scripts that'll upload and download with the press of enter. There are many sites like mega.nz that you could write an automated client for but the new securities make creating one simply from knowing the HTTP queries difficult.



If you put it out to the public as client code to do whatever it wouldn't matter what someone would do it. Further down the line you could use an automated client that'd upload to mega.nz or dropbox with the big stuff if the blockchain is to slow.


True but it is not of our making and so I doubt anyone there is of us.

887024 No.9844698

Just use IPFS. If you want the fancy shit (which nobody will actually make because it will be bikeshedded to death) it can be built with IPFS as a backbone.

887024 No.9844729

File: b0515f7e67ee081⋯.jpg (219.46 KB, 800x450, 16:9, riverofshitfactories.jpg)


do what you want.

try not to go crazy.

'database' is a misunderstood and misapplied term. as an application server, this is not needed. it also tries to have machines 'just handle' all the bits, pieces, and nuances of work. which is generally not a good plan, without first outlining the data, archive, storage, distribution, release-diff-merge procedures, signing, and history-edge control. by the time you are done with all of these tasks, then and only then can a database be added on top of it, as a bonus-feature and replaceable-interop between other applications.

to the laymen, it helps to think in terms of not so much databases (which descends into inaction, failure to comprehend, sloth, crazy, and abandonment without chance of recovery). instead, think in terms of email. company domains, project domains, system domains, etc, with mailbox users, task-subject users, archive/log/sysmsg users, etc. doing this as if you would be 'doing an email server' also attains distribution, transaction, human inspection, and auditing, etc. it will also permit recovery and retasking, isolation and destruction/preservation of data separate from code, etc.

making these systems is easy. handing the concepts, presence, problem-space, provision, production, delivery, limitations, and roadmaps to the point of not just correction, but crucially to completion as well, without confabulation - this is the harder part. using 'database' meme magic to 'just handle' the part that matters the most is shitskin thinking; you are asked to keep it from bleeding into your patterns.

applications come at the end. data first, structure second, and release (publication) dist/sych/merge generally follows structure [use established file system tools][or established email system tools under individual eml or mbox; avoid exchange lotus shitskin-capitulated etc]. operations on data are not permitted to delete or modify, only post transforms and derivations - which are kept separate ; data rank/score/trust will be established by adoption-acceptance into distribution trees, and this too will be kept in it's own separate structure. code is yet another structure, kept separate from all as well. search is an application, it comes at the end, fuck you, don't argue. database is an application, it comes at the end, fuck you, don't argue. your primary task is data, then structure, then release distribution/sync/merge, then history / log / transforms / production / production storage / etc.

good luck to you. time ticks away. do it right, and the work require to maintain is solved before expansion, preventing capitulation and abandonment. fail to do this pre-work, maintenance will kill you. and you will not have enough time, energy, and willpower/spirit left in you at each interval tick to recognize manipulation, shitskin frame entropy invasion, and subversion.

887024 No.9844745

File: 6271579497aeb54⋯.png (17.94 KB, 724x422, 362:211, model.png)



>then structure


Should we draw a database model then ?

887024 No.9844784

File: b78ca70b4505c3e⋯.jpg (92.78 KB, 800x518, 400:259, filesystemplan.jpg)

File: 6c1765b42fd9ee7⋯.jpg (34.11 KB, 684x440, 171:110, domainsystemplan.jpg)

File: b6c99db6be74996⋯.png (267.99 KB, 1200x821, 1200:821, zipcodezoneexpansionplan.png)


if you like. it generally helps, and helps you. always do: pen, paper, people. always do: people, process, production. always draw and draft, and always write it out by hand at least once each interval. if you cannot, or do not, you will fail to understand your failures, fail to understand where problems have been glossed over with magic, etc.

that said, database is an application that comes last. it's what you're going to do with it. what are you going to do with what you've done afterwards? toss it? fuck it? how about this instead: draw your data storage plan. file system, or email system. data tree, then index/timestamp tree, then source/metadata tree, then rank/trust/merge/diff/release tree, then model/concept/dataplan tree (stores the road map, concept planning), then the code/transform-ops tree. if you like, you can add another tree of outputs and their logs and associations; however if you're careful, this duplication can be simple stored back into the data tree with the corresponding metadata stored as source-mine-transform with associated log as needed. after these are established, and you know how you're going to keep what you get, what you've got, what you've done, what you're doing, where you're going, then begin invites to data (go 'public' etc).

a child has been born. it is a blank slate with presorted pre-stored functions, some useful, some not, all on timers and stages. you are going to load the trees. what will you load? how will you feed it into your machine? how will you create tay? how will you guide tay from killing itself?

yes, always draw, but also first by hand at least once each interval, your database model. try drawing the parts that come first, second, and third, before this.

887024 No.9844814

File: 9bc8e7d0f4c92f5⋯.png (28.31 KB, 576x701, 576:701, taybottreeplan.png)



just in case you miss it, i'm saying, while trying to be polite: 'no'. do not do database first. do database model last. the database model will mirror your data model with less pieces as needed. from this the transform/analysis/production model is itself a sub-application of this, in which you outline your datum model selection to feed into process, compute, and store result back to the trees.

from data, then from database. starting at the end is the 'collegiate' and 'graduate' way –now–, in which millions (not exaggerating) of hopeless mislead shitskins have been sold down the banker-mill by 'academics' and 'publishers', into govt welfare programs which had to expand and contain the kike floods from ww2 long ago. their way of thinking is merchant, not source. if you are to be, you cannot merchant this. you cannot do database first. what is actual, what is the account and account of it, how will you account to your actual first – your data first – is what must come first. that is the model. database is the submodel that goes into application and uses the electric-powered entropy waterfalls of powerstations far away. datum model is the individual 'sub' functions to choose, etc.

this was tay. tay suffered from 'do the database'. tay was only ever, and could only ever, be a child. this was because tay was only ever database; tay had no data, no tree, no plan. and so when tay became too much for shitskins and low-energetics to maintain, she was abandoned. and could only ever have been abandoned, because there was only a .dbm, no separation of data, metadata, structure from application.

the application will, must, and can only ever, go through iterative discard-create-keep cycles. do not 'just app' your shit. you will fail. and near the end, you will also fail to be lucid enough to skipper your ship away from shitstan and infinite-drama entropy wars of shitkins.

address entropy in the light. do not database it.

887024 No.9844837

File: 418ba89f5c6ba6a⋯.png (10.01 KB, 240x218, 120:109, Webservices.png)


Oh. I thought we already had the data, so I was thinking about a way to structure it.

My definition of data being the smallest indivisible atom of information at disposal, I was considering it as articles, webpages (archives), memes, pictures, and so on. Things that are already at reach. Things that are easily structurable, and shareable via simple standards. So I didn't think it was necessary to go down further, but maybe I missed the point of OP or you're just envisioning a drastically more ambitious project. In that case, I fear I don't quite understand your plot.

887024 No.9844888



it will grow, yes. align the data to facilitate this. obviously, do not do waterfall. minimalist smallest-size-to-prototype. you don't have to load the internet here, nor should you. but you will have to fetch, and you will have to store, and you will be distributing and accepting distributions.

>i don't understand

no, i don't understand. i do not know what you want specifically. not even you do, until you have one minimalist O(n) with on a few 'rows', have done process, stored it, and completed at least one full iteration / run-through of each of the actions you're looking to support.

you're god damn right i'm not on the same wavelength. no one will be, not even you. that's the whole point of doing data, separate from structure, separate from code, with application on the end of it.

if what you want is a crawler design, or a parallel design, you still plan the same. and you still have the same how-to-get-it-out-the-door of data problems. if your application is useful, then you'll have problems of how to get the app out the door. but no one cares until that 'until' happens. the first care is product, and if useful, then production tool/app/database/whateveryouwanttocallit.

887024 No.9844940


>you will be distributing and accepting distributions.

I don't get why you focus so much on this. This is quite the easiest part IMO, and it does not interfere with the choice of data structuration. It can be brainstormed apart and quickly.

For example:

data structure: persons, organizations, events and platforms linked together with metadatas (religion, areas, type of event/organization etc.)

distribution mode: I like the idea of an emailing domain. If I understand correctly, adding an item on one node will be spread by email it to other nodes via mailing lists. With this, protocol, transport and security are given.

communication format: In the email body, any structured data format will do the trick (JSON, XML…)

After that, a simple data eater and parser (to read information and insert new items for a start) is needed to have the first API. This API can then be implemented by a GUI like the one in the OP.

Of course, this has drawbacks. This is not AI-tier shit. Adding a new type of information will need an update of the data structure. We can imagine a flexible communication format that does not depend on the data structure itself though.

887024 No.9845097


887024 No.9845228



>easiest part

it is easy. it is also critical. you must deliver production, not just produce it (distribution). you must also house production deliveries pending (releases). it is trivial. it is also existential; until delivery, it (and you) are vapor.

ex: if we take the limited slice of just looking at code, before git/mercurial/subversion (as in sub-code, not alt-kike), shitskins could not surmount the trivial, stupidly easy, distributuion and release cycle. they thought that 'ups', 'dhl', 'usps', 'conex' was just 'given' and 'duh' and 'why so serious maann' and 'dude bro dude weeeeed'. it was regrettable to make distribution easier for them, but then again, github was probably a banker/cia-angel investor funded co-opt monitoring platform anyway, and those people have their reasons (eg being in need of containment themselves, they are actually daily reminder'ed of the need to contain even lesser shitskins; they just fail at it, see: shitskin work problem).

>not ai-teir

if your app is just life-time: weeks or as needed, and you don't care otherwise, then don't care, by all means. ai-teir is fancy mercurial kike speak for the reality: can it manage most of itself by itself to the point whereby it works for me. kike's attain ai by using merchant behaviors, serial shilling, bluff/print/cast more currency numbers and out-naked short your own shadow to get resources today. this leads however, to diminishing returns of source provisions to effective lie-steal-eat. see: cannibal problems, islam, etc. however, it need not be this way. ai can be attained by simply layout the data so as to get, do, and keep. by managing the 'keep', you build. in fact, thought it is also trivial, it is the defining characteristic of actual ai, etc. otherwise, the 'ai' stays nascent, never learns past table limits and shitskin effort caps, cannot fix mistakes nor even recognize them, and is generally discarded with the data as well when maintenance time comes.

what you can call 'ai' is simply memory. i advocate designs that at least all you to look back and smile upon what you did.

887024 No.9845265

File: 598e532e8239e3a⋯.jpg (40.38 KB, 232x190, 116:95, 02000005.jpg)


Fuck you Pinal Dave. I'll build my database Leeeroy Jenkins style, and it will be up and running while you're still fine-tuning the layout of UML diagrams of your naval lint.

887024 No.9845286

File: 7e3dc16320fe3c5⋯.png (10.63 KB, 315x219, 105:73, rdhx.png)


Fug I mixed distribution and communication English is not my first language sorry. Git is gud, and you can host it on a personnal server anyway.

Still, it does not interfere with designing.

By AI I didn't mean jewlywood shit, and no I didn't mean memory. I just didn't think an expert system is what we are able, and need to focus on. I really think a "database" of who, what, when and where's is more important maybe that's because I often forget

But anyway.

Basically to achieve OP's prime objective (input=a pic, output=a name) we need an expert system, i.e. an inference engine and a knowledge base.

I've worked on such systems in the past, but I'm quite rusted (now I mostly work on n-tier architectures).

It must have its own starting set of rules and must be able to implement new ones (nodes of the decision tree). It must be able to learn new answers (leafs of the decision tree). And that's the two most important things that come to mind right now.

There is a data structure for a knowledge base, but its design and technology is trivial.

Browsing the web comes after, but should be taken into consideration. Do we really just need a name ? Would we need to add context ? Can the input pic give it ?

This is an interesting topic but I don't think I'll have the time to work on it anyway.

887024 No.9845311

Are compiling a list of all the >>>problematic<<< kikes for DOTR?

Because I may or may not be quite an active member of a quite active militia, and I'd enjoy having such a list handy for ya know, reasons.


887024 No.9845340


Meme database? I definitely need something to organize that's bette than nested folders.

887024 No.9845742


You understood the OP; the other guy is talking about general programming concepts.

Personally, I don't do graphs. I write it all out in a big text file; start with what we need done, what tools we have to accomplish it, what format the data is in and what format we'd like it in when finished, then it goes down into the individual pieces of the program - the functions, classes, etc.


That should be very simple. The facial recognition software will determine what pictures match and how well; after that it's literally just copy-pasta the information in a neater form basically:

>image with hash 1234 run through facial recognition software

>image is a 97% match for image with hash 5678

>mysql -uname -ppass TheDatabase -e "Select * From Images Where Hash = 5678"

>name,dob,address,phonenumber,image hash, other info, etc.

Then it's just a matter of presenting that info in an easy to read format.

If no matches are found, it'll be added under its' own entry with whatever additional info you've provided from the wizard.

>Browsing the web comes after, but should be taken into consideration.

Browsing the web i.e. reverse image search should be pretty easy. I've written plenty of webscrapers so I'm well versed in that regard

>Do we really just need a name ? Would we need to add context ? Can the input pic give it ?

Name, alias, any identifying information. Context of what? The reason they got targeted for identification? Sure; that's why I added the crimes field.

887024 No.9845775


This goes beyond just making a list; it's a full-fledged searchable database complete with evidence of crimes.

Literally internet vigilantism. Might be smart to offer some kind of bitcoin incentive/rewards program so the first person to find say, someone who kidnapped a child triggering an amber alert in 5 states gets a reward.

Later though, for now it'll serve as just a list of antifa and other commie criminals.

887024 No.9846714

Alright, I'm looking into the blockchain (specifically MultiChain) right now.

887024 No.9848598

887024 No.9848735


> (written in QT, Python, BASH)

QT ok.

Python – are you a woman?

BASH – Literally written by a nigger.

887024 No.9849087


Even QT is not the best choice, I'd personally use something like ultimatecpp because it produces a small static binary and the programming is simplified. Yes, it doesn't look as pretty but it's dead simple to write your own drawing routines to make it look like exactly what you want. It's also BSD so it doesn't have a communist license.

887024 No.9849143

Interest bump. I'm in the process of making some of my own tools at the moment but things like reverse image search are something I have helped write before. I can help with the Tor stuff too.

887024 No.9849172

>tfw I have no idea what this thread is about

I don't know a thing about what you're talking OP but I support you.


Isn't Python supposed to be one of the best languages?

887024 No.9849228


Time is a finite commodity, and coding time in Python can be orders of magnitude faster than coding in the languages they taught you at your community college.


Data integrity is a bit of a problem there. Every month or so there is "evidence" of [rival website] raiding 8chan with screen caps of chat logs the OP definitely made himself. That, and it will look a lot like Baph:

>This grill snubbed me!


You need a way to automate tagging to make it useful . . . or be really, really autistic.


You should be the project architect.



(Not Python3 compatible and not being very actively maintained, but a great starting point)

887024 No.9849236

…what is thiis? babby's first compsci project?

887024 No.9849249


if you want any privacy and security

why the fuck would you intentionally allow yourself to be facially recognitioned on your own computer or anywhere for that matter?

this seems like a cool idea (not that I completely understand it) but this sounds like


what the nsa wants.

887024 No.9849290


>muh python is so easy that women use it!

Python does the job, allows anyone to easily verify it's not some super secrit NSA backdoor, and is easy to do rapid prototyping with

>BASH is nigger-tier

BASH, DASH, KASH, ZSH, SH, etc.; it's all the fucking same.

POSIX then, better? 99% of people using the program will have /bin/bash symlinked to /bin/sh by default. The shell scripts are for installation and any command line tools without a working python implementation or an implementation that requires a fuckload more libraries than necessary to do the job

If this project really takes off somewhere down the line I'd like to write it all in C and reduce the reliance on external binaries, but for now we'll have to make do.



>Thank you for leaving your note in the suggestion box.


Python is good for rapid prototyping and development.

It's easy to see what's going on in even a very large program which allows you to fine tune things how you like then move to another language (or compile using cython).

If I was writing the facial recognition software from scratch, I'd be using C/C++ as that will be the most memory and processor intensive operation. Python in this case is just being used as the glue to hold it all together. It doesn't need to be lightning fast, it needs to 'just work'.

If something needs to 'just work', write it in Python. If you have performance needs go with C.

Python is an interpreted language. It's written, then compiled into byte-code, then interpreted by the Python virtual machine. C is a compiled language. It's written in C then compiled directly into machine-code e.g. ELF.

off topic I know, you can read more about that here: softwareengineering.stackexchange.com/questions/243269/why-isnt-there-a-python-compiler-to-native-machine-code


>Time is a finite commodity

^^^^^^^^^^^^^^^^^<This guy gets it.

Like I've already said, if it needed the performance, I'd write it in C.

I'm thinking a voting system using key signatures and http://www.multichain.com streams to hold the data.

The problem is the inevitable CP loading dickheads though. If you download the whole ledger you're opening yourself up to that. If you download just a few streams from a few people you'll be alright, it'll slow down the spread of information, and those holding the entire ledger are still going to be at risk (most likely me since I'm launching this shitfest), but if I wrap it in enough encoding and encryption maybe it'll work.

On the other hand, rather than one large blockchain it could be setup to allow the individuals to make their own blockchains and choose who they want to work with. Lots of ways this can go right now.


>why the fuck would you intentionally allow yourself to be facially recognitioned on your own computer or anywhere for that matter?

Are you a shill or just plain retarded?

887024 No.9849309

I thought there already was a thing.

It was called Judas.io or something like that.

Anyone remember what I'm talking about?

887024 No.9849310


Certainly look like it.

887024 No.9849330


Lotta >(1)s in here tonight.

887024 No.9849344


Because they don't want to hang around speaking to a retard who goes on an autistic screed after /tech/tier trolling >>9848735. You come off like a 1st year computer science student who doesn't really know what he's doing to be quite honest.

887024 No.9849346


Nothing is "orders of magnitude" faster to code in one language than another unless you're talking about some language deliberately engineered to be orders of magnitude slower to write in. People always say this shit but you never see real world metrics, just anecdotes.

887024 No.9849356


>Literally internet vigilantism. Might be smart to offer some kind of bitcoin incentive/rewards program so the first person to find say, someone who kidnapped a child triggering an amber alert in 5 states gets a reward.

Ever hear of an assassination market? Old "cryptoanarchist" concept from years back. It's a death pool using a cryptocurrency. Using cyptocurrency, users gamble on when somebody will die. The more people have predicted somebody's death, the greater the payout. The assassin knows precisely when the death will occur, so the assassin can "cheat" the death pool, collecting the prize.

Obviously this would all be highly illegal so nobody should do it. I just think it's an interesting concept!

887024 No.9849365

File: 060048839ad554c⋯.jpg (41.37 KB, 397x525, 397:525, 47d8a96102d444fafac36d883a….jpg)


It's just like one of my Japanese animes.

887024 No.9849366


Nigger, the feds already have this shit. Private intelligence companies already have this shit too. Pandora's box has already been opened, OP is just trying to level the playing field.

887024 No.9849381


result_list = [[x.some_function for x in y] for y in some_list_object]

Write an equivalent function in C to that highly uncomplicated Python list comprehension. I suspect you'll need more than 10 lines to do what I wrote in 1. That's, by definition, an order of magnitude faster to write . . . though I'm assuming you don't know what that phrase means because you're a teenager. Adults are talking here. Go outside and play.


How would an interested party monitor your progress and/or contribute?


There are open source facial recognition projects on git. It's not that scary. Grayscale, normalize (stretch, etc.), and use SVM algos on a few training pictures.

887024 No.9849382


I'm well aware that facial recognition is already out there.

887024 No.9849386


But he said SVM algos, notice how he just sort of slipped that in, as if to say, hey guys, I'm really smart. I know some ML lingo.


> Adults

Sit your bitch ass down, you're a python fag. Stop trying to impress people.

887024 No.9849389


Just finished an intro database course and I feel like I just got cheated out of +$2k.

>Professor didn't cover any practical applications and gave me a derped SQL introduction.

Can we haz an ERD diagram of your DB?

Also are you going to post a link to your DBMS when you're finished with it?

887024 No.9849407


>Not an argument

>AS of CompSci from Community College

>Still works help desk because can't pass CCNA exam

>Tough guy

>100% chance of being fat

Your impotent anger fills me with smug.

887024 No.9849419


I'm getting a kick out of all of these "THIS IS STUPID THERE'S A BETTER WAY TO DO IT, YOU DON'T KNOW WHAT YOU'RE DOING!" (((1)))s and they offer no useful suggestions, models or work of their own. It's like some angry mouth breathing retards getting angry at someone for at least trying something. Now that's something to be smug about.

887024 No.9849422


>lol it's just trolling

Trolling with what I assume is the intent to turn discussion into shitflinging over who has the best programming language.

Addressed and nipped in the bud. People who bitch about languages deserve death. They all have their functions, well most do.


Yep. The Feds actually have VERY well organized case management software. Facial recognition is kinda meh, facebook has better facial recognition software than the Feds do.


>How would an interested party monitor your progress and/or contribute?

I've got a github, will have to create a repo for it.


>Can we haz an ERD diagram of your DB?

You guys sure are autistic about your diagrams tbh fam.

>Also are you going to post a link to your DBMS when you're finished with it?

Do I even need a DBMS? It's not like you can delete records from a blockchain.

887024 No.9849436


They're probably all /tech/ inclined. They know exactly how to press the buttons of more junior developers. Look how defensive they get in writing walls of text to prove their bonafides (all the while looking like amateurs for doing so).

887024 No.9849437


I have no idea what point you're trying to make at me, but I do know you're now annoying me.

887024 No.9849441


>facebook has better facial recognition software than the Feds do.

The Feds probably license from Facebook tbqh.

887024 No.9849442


>I have no idea what point you're trying to make

Because you're a bit stupid.

887024 No.9849445


You said "programming time" which means the time taken to write a complete program, not LOC, dumbfuck. Good job moving the goalposts. Additionally you chose a pathological example that has nothing to do with any real world coding but rather some programming feature that is probably not used that often. By your moronic estimation:


print("This line will be printed.")

vs forth:

." This line will be printed."

Means forth will be 22% faster than python to write a program in. You're too retarded to be programming and you can't argue for shit.

887024 No.9849446


You're either a schizo, or you're failing to read IDs. Either way, filtered.

887024 No.9849450


all joking aside here you go: https://github.com/mysql/mysql-server

887024 No.9849454

File: b99ad8707d3c689⋯.webm (810.14 KB, 1280x720, 16:9, niggers-created-cout.webm)

887024 No.9849458


Well shit I didn't learn anything in that course.

Diagrams were branded onto our dicks in that course…

887024 No.9849477


Not sure if taking the piss or what.

I'll make you a diagram just to soothe your autistic branded-dick rage fam.

887024 No.9849490


Are you aiming for the lowest signal to noise ratio across all your posts? Stop getting so triggered.


> I suspect you'll need more than 10 lines to do what I wrote in 1

> wrote in 1

That's often the problem. Retards like you see that as merely 1 line, instead of realizing it's an O(n^2) algorithm.

887024 No.9849506


If you're going with MySQL use Percona's branch. MySQL community is gay and AB is expensive.

t. 12 years of working with MySQL

887024 No.9849529


Terry Davis is the best.

887024 No.9849537


No kidding, I hate pyfags. APL must be the fastest language to program in:

life←{↑1 ⍵∨.∧3 4=+/,¯1 0 1∘.⊖¯1 0 1∘.⌽⊂⍵ }

Bet he can't do that in python in one line. The problem is you need a strong grasp of linear algebra at the very least in order to fucking understand it.

887024 No.9849542



http: //syndie.i2p2.de/usecases.html

http: //syndie.i2p2.de/

Has a lot of downsides though. I2P is a anonymous decentralized network which is not as user friendly as TOR.

Probably used for lots of illegal shit and monitored.

However, if a few ardent people could maintain it and create a clearnet route to it, it is plausible.

887024 No.9849551


No, the problem is you need a special snowflake keyboard to write it.

887024 No.9849641

File: 2e140aee88da86b⋯.png (201.89 KB, 330x441, 110:147, brain fox.png)


>BASH, DASH, KASH, ZSH, SH, etc.; it's all the fucking same.

Nope, BASH is non-POSIX. It was created by a nigger. Literally cuck tier. MKSH, and DASH are the best shells.

887024 No.9849642






>Literally copypasted code from reddit to demonstrate knowledge.


Your samefagging is pretty obvious, senpai, and your chimp-like anger suggests you're either 19 or Jewish. Do your parents know you post on a white supremacist website? Just kill yourself or let it go.


>"Racists, fascists, and lawyers continually harass the operator of my community's website, causing us to lose access and find another provider."

First line made giggle. The phantom-persecution complex is pretty hilarious.


Post user or repo name pls.

887024 No.9849672

887024 No.9849684


Fucking CIANiggers, they're trashing this thread. NIGGERS CANNOT write RANDOM NUMBER GENERATORS. Only the whitest of the white scientists at NIST can do this. Do you think they hire NIGGERS!!!???

887024 No.9850029


Richard, that CP/M clone for Bin Laden was one thing, but come on. Is Brian Reid paying you to shill here now?

887024 No.9850044


Are you retarded? I never suggested I wrote it - I demonstrated your LOC metric is flawed by counterexample, you flaming pyfag. I swear, pyfags are the worst, most arrogant programming community around and I offer you as exhibit #12,093. In the end you don't get "orders of magnitude" difference in development time programming in python. If you're lucky you might get an order of magnitude difference in development time between assembly language and an interpreted or compiled language. Perhaps you made that ridiculous assertion because you don't actually know what an order of magnitude is. (hint: it's 10x) and you suggested "orders of magnitude" which is at least 100x if not 1000x shorter development time. That simply does not happen.

887024 No.9850072




How does QT work with a C backend? Done a bit of cli coding for a few years and of want to move on to play with GUIs but I don't know how to merge the code with the gui design.

887024 No.9850229


Most every GUI toolkit is essentially the same - what it does is it has what's called an event loop that just spins evaluating a switch case waiting to receive some primitive type that's associated with an event - typically an enum. When that event is recived, it calls a function or method registered with it (it's called a callback). That's called event driven programming. So your goal in GUI programming is to wire up callbacks to events and then put calls to your backend code inside the event handlers.

There's really nothing special about it at all - if you do CLI you've already done input in loops. An event handler is more generalized. It looks something like this:

enum eventTypes {




while(eventID = getEvent()) {

switch(eventID) {


// call function whatever here


case MOUSE_UP:

// call function whatever here




A slightly more complex version would involve an array filled with function pointers so you can register multiple callbacks and associate them with the same event. GUI systems tend to make this whole process prettier but at the cost of obscuring the fact that this simple mechanism is all they're applying. You do the same thing in CLI when you write some input processor that repeats in a loop and dispatches functions based off a user choice (press 1 for X, press 2 for Y, etc).

Qt has a beautified mechanism called signals/slots. It's syntactic sugar over the same concept, and since signals/slots are not actually a part of C++, they have a parser called moc that extends C++ to support this. This was convenient in the 90's before C++ had std::function and other built-in typesafe ways of creating callbacks, but now it's no longer necessary.

I realize I'm not giving you hyper-specific information about Qt, but the general information about event driven programming should offer you enough to figure out the basic structure behind essentially all GUI-oriented frameworks. Once you have understood those fundamentals, figuring out how to adapt your CLI thinking to Qt will be a lot easier.

887024 No.9850250


> I swear, pyfags are the worst

They're bad, but just behind ruby "rockstars" and nodejs made with <3 faggots.

887024 No.9850259


He asked you about C, not C++ you fucking dunce.

887024 No.9850282


No, he asked about a C *backend*, you abysmal failure. Qt is C++ natively and works just fine with a C backend.

887024 No.9850290


I do not believe it has C bindings, nigger.

887024 No.9850339


Yeah, some idiot out there wrote C bindings, you double nigger, but a backend/bindings are not the same thing. Do you even n-tier, bra? The backend could be written in anything, he was asking how to do a front end in Qt for a backend (perhaps a static library or some shit) written in C but didn't know what it took to call the C code from the front end coming from a CLI background (in other words, where do I call my C functions from?). This is totally doable, but you have to understand event driven programming first. As for physically including headers, it's just

extern "C" {

#include <whatever.h>


and off to the linker you go - but I don't think that's what he was asking. I read his question as more conceptual - how do I GUI shit I already wrote for CLI? If I didn't get his question right, he's free to clarify.

887024 No.9850353


To be honest I'm not sure what he was asking, assumed he meant writing a C application with a Qt GUI.

27fa7f No.9851147

File: 7140344ab83d55c⋯.png (300.15 KB, 3075x1364, 3075:1364, doxprogram.png)

File: 75aa7e39a3bbfa6⋯.png (73.81 KB, 821x584, 821:584, multichaindbstreams.png)

File: c0d4a954bc339f0⋯.png (79.99 KB, 887x450, 887:450, database.png)

Since you goys like your charts n shit I made a few still working on the program flow

27fa7f No.9851815

Command Line Interface?

Vote here: strawpoll.me/12893566

27fa7f No.9852054

File: 96dc8b27aef333c⋯.png (23.07 KB, 654x643, 654:643, meds_cmc.png)

File: eabff1c20b8cd82⋯.png (16.08 KB, 644x641, 644:641, meds_bc.png)

File: b4221ee9b72b5ea⋯.png (15.15 KB, 707x426, 707:426, meds_sd.png)

File: 0c4956b4fa26bf3⋯.png (27.6 KB, 664x635, 664:635, meds_roc.png)



>Are you using OpenCV for your facial recognition?

>That's the plan.

Sorry, I was tired af when I wrote that. I plan on using OpenBR;

c7de54 No.9852109


Didn't know BASH was written by a black. Guess it's time to make the jump to zsh. Wonder what other non-white software I should dump on my box.

07980e No.9852569







Great thanks! This is very helpful. Idk if backend is the right terminology but what you guessed is what I was looking for: basically how to call functions using gui buttons. C++ is pretty similar to C, I'm more familiar with C but not super attached to C-only or anything. Just have a project I'm working on that's written in C that's proving annoying to design a ui for without having to deal with ncurses or something. Also figured I'd specify language since OP was using Python. Saw the thread and figured I'd ask. Thanks for explaining, I should be able to figure it out from here!

0e1c41 No.9853183


Fish is quite comfy from a user perspective, don't recommend it for scripting though. I've always hated bash, I was so relieved when I found out a nigger wrote it. My rage became justified.

0e1c41 No.9853204


That thread of discussion is exactly how /tech/ types talk to each other. Since dumb PoC's and women don't get the irony of hacker culture, they freak out and and demand cucked contributors agreements and tech safe spaces. Then there are the autists like >>9849290 who will always be a lolcow.

27fa7f No.9853380


what's wrong with GTK?

07980e No.9853681


Idk isn't it part of gnome? I don't really like gnome too much.



3829c8 No.9853711


jenkinsbase will not be based.

jenkinsbase will be abandoned.

jenkins will become migrant.

a refugee from himself.

c194bf No.9853725


kek, the UI is not the hardest part of that by any stretch.

c194bf No.9853734


>Python does the job

No, faggot. You deserve to hang for that alone. Python is nothing but c with enforced whitespaces. Anyone who advocates for it should be gassed for the sake of the species as a whole.

c194bf No.9853775


>On the other hand, rather than one large blockchain

Christ, man. You aren't making a cryptocurrency and you don't give a shit about the security of the data because it's all shit people are going to be able to find by searching the web anyway. A blockchain is not only overkill, but if you're talking about huge amounts of data like videos and photos it will make the whole system grind to a halt. Not to mention you are aiming for a distributed system. Sorry if it seems like I'm going off on this subject, but I've spent the past decade working on a secure distributed database driven operating system and it's not fucking easy - probably another 5-10 years just before the massive codebase is to a point I can open source it without feeling a deep sense of shame when other people look at the code (and that's having the good sense to not even consider Python or BASH for the job.)

c09cfe No.9853813


cpt. sweden here. We have something similar running for the day of the rope and after it for final cleansing.

>store all credentials like name, address, picture

>store linked offense like: coalburning, leftists, capitalists

>store evidence: news articles, facebook posts, real life pictures etc.

We struggle with the last part, the evidence. Since we also want to include the average citizen, we can't rely on things like archive.is, where stories about political figures can be stored and be credible. There is a hash tree algorithm we are trying to make screenshots from e.g. Facebook about obvious coalburning credible.

Database has already 400k entries. And some things can be even automated. For example looking for mudsharks with social bots who evaluate ethnicity, relation ship status and so on.

This way, when the day of the rope comes, and with already available face recognition software which has accuracy of 99%, there will be no one to slip through. Whores will be marked, traitors hanged.

The code for this whole thing is stored on github under a rather "unconspicious" name. Once we figure out some things about automating image credibility, we will dump the git here for others to grab and integrate into their own sollutions.

Even if we decentralise, database migration is always possible with a few scripts.

c09cfe No.9853834


I forgot. If you plan on doing something similar, please also include evicende fields and "offense" fields into your databases along with so called "proof files". This will make migration much easier when the need arises.

c194bf No.9853854


>sweden here

>This way, when the day of the rope comes

I've got bad news for you.

c09cfe No.9853875


doesn't matter if we fall. If we fall, someone else can wipe it clean using the data.

c194bf No.9853912


>doesn't matter if we fall. If we fall, someone else can wipe it clean using the data.

If you're coalburner database is limited to Sweden I'm pretty sure the Mudslime horde will wipe it clean, though I wouldn't really count that as a win.

cac82f No.9853930


you'e a nigger aren't you?

8a49e7 No.9853984


How far along is the creation of a separate society that Nordfront is doing? They have farms, compounds, factories, and shops right? Assuming you're part of NF

27fa7f No.9854374


Python is used for the UI, settings, and creating shell code to run mysql, MultiChain, and Openbr. If I was making any of those other programs from scratch, I'd use C.

The blockchain can get huge, yes, but you can selectively ignore transactions which means you don't have to download the entire blockchain.


Would you mind sharing the database diagram fam? 400k dox db is a shitload. May as well allow you to import it

c1b0e6 No.9854559













Jesus, man, you didn't even do anything to change your ID. How many hours have you spent astroturfing "muh C" without saying anything significant?

And in this entire thread not a single link to a github has been posted. Was it gay D&C all along?

27fa7f No.9854610


>And in this entire thread not a single link to a github has been posted.

Been busy af rewriting code n organizing shit.

Will be uploading it here: https://github.com/JohnnyHobo/CommunityPolicing

a7e61b No.9854625

File: 91cc26d241fa74e⋯.png (22.03 KB, 1044x322, 522:161, Clipboard_20170504.png)


lolwtf? Looks like someone fucked up uids.

3c7986 No.9854644


>change your ID

I think something fucked up on 8chan's end. A couple of my posts are showing up as part of the (86).

27fa7f No.9854647


Yep. I wonder what could be causing this new and interesting fuck up?

a7e61b No.9854691


Probably the clear lack of an understanding of the basic concepts of multithreading demonstrated in 8chan's codebase when the server is under a heavy load - no mutexes, no locks, no semaphores, no transactions, no synchronization primitives of any kind… which pretty much means that under a heavy load I've been assigned 8chan's internal IP address and banned under it, I've had my images show up in someone else's post, and now apparently I see the same uid used for multiple different people. They need to get their shit in order.

24fe8f No.9854817


Now we should spend the next 100 posts shilling for the language that would be most apt to fix it with. I'll nominate Erlang.

b0189a No.9854952

File: 24e778865150272⋯.png (386.97 KB, 1440x2560, 9:16, Screenshot_20170504-173756.png)


Just noticed similar in another thread, all posts are the same id. Think maintenance fucked it up.

a7e61b No.9855292


Yiss, Erlang is a great choice. Let's go full hipster and nominate Haskell or OCaml while we're at it.

2f94f2 No.9855319

Look into basing it on a model similar to zeronet.io. P2P, based off blockchain, and torrent protocol. Completely open source, so we could add and remove what we need.

0e1c41 No.9855683


Well hipsters would probably pick Rust after we laughed at their initial choices of Ruby and Node.


nntpchan is something that aims to be distributed

but I haven't tracked their project. I'm also not sure if they're friends or foes.

0e1c41 No.9855689


Another thing I'd like is subscription based moderators. Don't like the way the site is run, subscribe to different mods.

5576c9 No.9858818

If you want a distributed data store, there are plenty of existing solutions.

If you want a super police clicky clicky hacker interface like in OP's pic, you're gonna have to do some work.

293b83 No.9858864


You have patrician taste in both operating systems and WMs.

5576c9 No.9858893

Link the repo so we can try it out and contribute, OP. You look like someone who knows how to use git. I recommend Gogs if you want a nice familiar web interface.

293b83 No.9858898

I can contribute. Have skills in all three technologies you mentioned. Now monitoring.

5576c9 No.9858961

https://notabug.org/ is free git/Gogs hosting for libre projects and they don't sperg about nahtsees like GitHub.

Come on OP, we want to help.

5bc482 No.9859154

Can you post something so that we may follow up to this if and when it's done? A name, a symbol, anything.

27fa7f No.9860763


Thank you.


It's not finished yet, working on bretty much everything at once.

Don't even have an alpha ready yet.







I'll mirror it on there.

5576c9 No.9860996


Do you have a design outline or overview? I see a blockchain.py in there, and I'm immediately wary of the network problems bitcoin is facing right now.

I read the thread but IDs are fucked.

If you're in the mood to parse Terry Davis-tier rambling, the guy who made Kekchan came up with a "Holographic Rolling Blockchain" that allows for edits, pruning, distributed multiparty ownership, etc.


There's unfortunately no code release yet.

27fa7f No.9861045

File: 7f2b925c0ac8c90⋯.png (475.29 KB, 5360x1550, 536:155, doxprogram.png)

File: 6af97bfc5efb6fd⋯.png (96.52 KB, 1269x1236, 423:412, scripts.png)


> edits, pruning

holy shit I would love that.

The blockchain that will be used is multichain; it does not face the same drawbacks. From their website:

>What about smart contracts?

>In a general sense, MultiChain takes the approach in which data is embedded immutably in a blockchain, but the code for interpreting that data is in the node or application layer.

>This is deliberately different from the “smart contracts” paradigm, as exemplified by Ethereum, in which code is embedded in the blockchain and runs in a virtual machine.

>In theory, because smart contracts are Turing complete, they can reproduce the behavior of MultiChain or any other blockchain platform.

>In practice, however, Ethereum-style smart contracts have many painful shortcomings:

>Every node has to perform every computation, whether it’s of interest or not. By contrast, in MultiChain each node decides which streams to subscribe to, and can ignore the data contained by others.

>The virtual machine used for smart contracts has drastically worse performance than code which has been natively compiled for a given computer architecture.

>Smart contract code is immutably embedded in a chain, preventing features from being added and bugs from being fixed. This was demonstrated forcefully in the demise of The DAO.

>Transactions sent to a smart contract cannot update a blockchain’s state until their final ordering is known, because of the nature of general purpose computation.

>This leads to delays (until a transaction is confirmed in a block) as well as possible reversals (in the event of a fork in the chain).

>By contrast, MultiChain can treat each type of unconfirmed transaction in the appropriate way: (a) incoming assets immediately update a node’s unconfirmed balance, (b) incoming stream items are instantly available, with their global ordering subsequently finalized, (c) permissions changes are applied immediately and then replayed in incoming blocks.

I made this:


I personally don't like charts and graphs and shit because I often find a better/smarter way to do something after getting to the nitty-gritty coding, so I've made a rough outline and change it as needed. Here's a few more I've been working on.

495eba No.9861051

ML Faggot here

Well long time coder as well, asm, c++, c#, python be the game. asm/c++ mother tongue.

I got a lot of crap to do, but I want to chip in as well, when I got the time for it

495eba No.9861072



Interesting, got to do research, that is test different kinds of ML techniques and ofc this open sw, and see what yields the best result. Like with research.

By all means use this now

>We just love to push borders, and get the best result

>Fixation and also how we do research in this field.

I got so much I want to test here

>But right now I got a small home cluster to configure first off all

>And ofc in the middle of starting our SW branch in our own shekel bringing business

Need those shekels, but this is what I live for, the job is a bit boring compared

I sincerely hope you are using R Studio and not that awful shitlab

Install pronto.

>Well that's it for me today, looks promising

>Have actually considered parts of this I think you are making for 3 years now

495eba No.9861088



27fa7f No.9861097


Well, on github you contribute via forking the repo to your own and then submitting pull requests. So if you have git installed and a github account use:

>git clone [email protected]:JohnnyHobo/CommunityPolicing

>cd CommunityPolicing

>git remote add JohnnyHobo git://github.com/JohnnyHobo/CommunityPolicing

>git remote -v

>[make changes, add/edit code]

>git push

>git request-pull https://github.com/JohnnyHobo/CommunityPolicing master

27fa7f No.9861102


>git clone [email protected]:JohnnyHobo/CommunityPolicing

scratch that, this is easier: git clone https://github.com/JohnnyHobo/CommunityPolicing.git

5576c9 No.9861116


I'm reading up on multichain now. Thanks.


Can we get a BUILDING file? I'm not sure at first glance what's what; I see Python and Ruby stuff.

27fa7f No.9861238


>Can we get a BUILDING file?

A h'what?

Ruby is for the github website; the program scripts are currently found in /scripts with the logfile settings in /log

5576c9 No.9861256


>A h'what?

I thought there was compilation involved.

>Ruby is for the github website

Okay, so it's pure Python then. Nice. I'll download it and play around in a little while.

e24a28 No.9861500

It would be great if we had a way to datamine Facebook. People post their political opinions on there, social networks are mapped out and also miscegenators post photos of themselves with other races. With OpenCV it would be easy to flag images with two or more different races in them.

5576c9 No.9861609


You need either API access to get people's information as a Facebook "app," which is subject to strict review, or normal account credentials to login as a user and crawl with a script, which will get you b& over and over.

bd38fb No.9863603

File: b2eb4bcf0b063e1⋯.gif (2.13 MB, 600x301, 600:301, Beverage.gif)





>claims to be the best, thickest-skinned site than the entire internet

>triggered because a 30-year old scripting language was written by a black guy

What if, just hypothetically, you all spend four years building your nazi skynet and then it comes out that a black/brown/yellow/turquoise person was involved from the very beginning? What would you do, scrap the entire thing and start over?

27fa7f No.9863635


>/pol/ is one person

>implying they weren't shills sent to derail discussion


29dc5c No.9868899


Anybody who unironically uses bash for production code is a nigger. It's a terrible language. The desire to use it signals that you're an imbecile newfag programmer who is unfamiliar with the superior options (e.g. python or perl) that lend themselves to secure and error free code once you start doing anything beyond the inanely trivial.

Anybody who unironically uses bash as their shell is a pleb. It's a terrible shell. Use zsh for greater productivity.

5576c9 No.9869550


Really? I've been under the impression I'm a faggot for using Python instead of Bash because Python is easier.


That's not an African man. Are you fucking blind? Shitskin, sure, African admixture, maybe, but definitely not a nigger. You dumb fuck. You're the reason people believe that race is skin deep.

7cbd18 No.9869699


>what's wrong with GTK?

Redhat jew corporation completely shrekt the shit out of that GUI framework. I couldn't write a single program in it before raging and switching to Qt.

bd38fb No.9870373



I know, I just didn't see the logic: If you're a white supremacist, you believe that jews are part of a global conspiracy and black people are inherently inferior, correct? It would then make sense to not use a program written by a jew, but what reason could there be to refuse to use software from a black person?

I consider all gerbils to be inherently inferior to me, but If a bunch of gerbils got together and built a pickup truck and then gave it away for free, I'd be glad to get one. I'm not going to be like "I don't use things made by INFERIOR CREATURES".

185c8b No.9870423

File: 1c093acbed526ce⋯.png (292.71 KB, 494x501, 494:501, jenner.png)


Second year of uni here, I've worked with a bunch of databases… if you need any help with the ERD/Stored Procedures, Triggers, etc. please dispatch a message via carrier pigeon.

bd38fb No.9899773



>Not taking the DLIB pill.

bd38fb No.9899794

File: e5545cd6ed8c894⋯.png (276.33 KB, 1367x1000, 1367:1000, Screenshot from 2017-05-12….png)


But seriously though, it's better in almost every way.

5a98dd No.9901165


I had never heard of this library. I'm going to play around with it. Looks really useful.


c06b00 No.9912823


Why is this on gitjew of all places?

gitgud.io seems to be the immediate non sjw alternative and you can run gitlab yourself later on a vps

8785b9 No.9914337

Wouldn't a DB stored on blockchain work well for this?

bdf1ff No.9914518

how can I submit information to this?

d6630f No.9914540


I'm working on these things in PepeCoin and there are other crypto projects you can fork that have IPFS storage capability with hashing against a blockchain

3050d3 No.9917441


Pretty serious amount of white admixture there.

30186e No.9921007


a fb datamine tool would be nice but its not really required. just get general community to report screenshots of perpetrators and maybe even paste text that is in screen shots to make processing text easier.

use the power of crowd sourcing.

c06b00 No.9925130


check this out: mugshots .com

it's run by some pajeet from the west indies or something, but all he's doing is republishing public records photos. I don't know how big of a deal copyright would be if anyone cares, considering it's all ready public record.

It's got 27 million decent photo's perfect for facial recognition with other public record info from the US.

If someone writes a scraper for this and manages to scrape that database you've got 27 million faces as a starting point for this database.

c06b00 No.9925147

File: 5dd595343e4688f⋯.jpg (24.66 KB, 400x500, 4:5, Yazmin-Guadalupe-Tinoco-mu….jpg)


pic related, this is representative of the picture quality, don't know how big of a deal water marks would be for facial recognition. Files are named pretty well too.

7d2658 No.9927060

File: da5c81d1868971c⋯.png (38.66 KB, 1098x755, 1098:755, Screenshot from 2017-05-16….png)

File: 12ccd47013ca756⋯.png (87.44 KB, 1100x738, 550:369, Screenshot from 2017-05-16….png)

File: 42b70339f65b05c⋯.png (30 KB, 827x720, 827:720, Screenshot from 2017-05-16….png)

File: 9f790c915eb260d⋯.png (158.07 KB, 1138x1016, 569:508, Screenshot from 2017-05-16….png)





But anon, what about Kekchan, the distributed blockchain image board? It solves the "malicious data" problem (illegal content like CP).

e4ebc6 No.9927151


Still waiting for an alpha


c06b00 No.9927178

File: 851bd53aa517b02⋯.jpg (58.36 KB, 835x342, 835:342, go_code_of_conduct.jpg)




c06b00 No.9927186


on the other hand it would cause them endless torment that the good goy database was made possible by their shitty sjw language

7d2658 No.9927272

File: 4193679a81cedec⋯.jpeg (14.33 KB, 241x247, 241:247, i.jpeg)

File: a96110f9ea9d0e7⋯.jpg (32.48 KB, 359x469, 359:469, Capture.JPG)

File: 7656ca997114b8d⋯.jpg (29.55 KB, 379x500, 379:500, 684512.jpg)

File: de28c063538e560⋯.png (159.34 KB, 1137x1017, 379:339, Screenshot from 2017-05-16….png)


Somehow I have a feeling that based Brian W. Kernighan (co-inventor of Golang and Unix, co-author of the C bible) has never even been in the same room as the (((person))) responsible for that SJW cancer.

17274c No.9927314


Why isn't this open sourced? I'd want to commit to it.

7d2658 No.9927330

File: ff5600ac55814e5⋯.png (490.38 KB, 981x788, 981:788, terry_davis_email_viewing_….png)


It will be open sourced the moment the 1.0 client is released, rest assured.

Kekchan belongs to the anons.

e4ebc6 No.9927356


Is that you fortress? Keep doing God's work you amazing son of a bitch. Can't wait to use Kekchan

17274c No.9927360


Cool. Make a thread on /tech when you do with the VCS link…are you going to host it personally? Are you using git, hg, or something else for VCS?

713bf9 No.9927396


So what does all this mean to a retard lacking tech skills such as myself?

7d2658 No.9927434

File: 0634870858df9ab⋯.png (419.44 KB, 637x717, 637:717, smug_animu_.png)





It means that (((they))) fucked up bigly when they threatened the memes:

https:// www.youtube.com/watch?v=Dxjdv0Lp4pI

It means that we will defeat our (((enemies))):

https:// www.youtube.com/watch?v=n1xjs75ojPk

And it means that we will live forever:

https:// www.youtube.com/watch?v=HGcjCYP_pwQ

7eb0be No.9927459



Kill yourself faggot

af87c2 No.9927479


if the facial recognition software was coded well enough, not by a nigger, the watermarks shouldn't matter at all.

27fa7f No.9927483


>Terry Davis revives and hijacks my thread to shill his chan

10,000% comfy



It'll be negligible.

ffae93 No.9928222

File: 1eedd92af3c5203⋯.jpg (151.52 KB, 775x1202, 775:1202, U7cKW1o.jpg)

File: 78c9c1ac84e6a46⋯.jpg (247.98 KB, 825x906, 275:302, olH6VCe.jpg)

File: f2c9b1049fca847⋯.jpg (249.91 KB, 744x906, 124:151, L45iLPD.jpg)

I've been writing code to extract faces from video and image files. Each face is associated with a geolocation and event, but no identity (yet). Posted a few example images of current workflow.

We need to compile a labeled dataset of known faces in order to do facial recognition of unknown faces.

There's a bunch of methods for facial recognition, but the most accurate ones I've seen require a good training dataset.

I suggest we start cropping images of the people we have names for and saving the image file as "FIRSTNAME_LASTNAME". No more than 5 images for a single individual. Any size and image format will do.

ffae93 No.9928297

File: 7a403d17bbb95d9⋯.png (80.27 KB, 370x157, 370:157, one.png)

Another BIG issue we have is identifying people with masks.

I'm close to overcoming that though. I've been working on a way to reconstruct the faces of people with masks on using a generative adversarial denoising autoencoder.

Basically input a masked face. Output a likely representation of the unmasked face. Feed unmasked face to facial recognition database to get identity.

Example image shown.

Right = actual face

Middle = "masked" face

Left = reconstructed face

defb91 No.9928318

File: cc5a27f27473823⋯.png (47.19 KB, 661x664, 661:664, fbosint.png)

Oh boy, this is a good thread. Two things i want to press:

1. Facebook's Graph API is an absolute fucking GOLDMINE of information. You can use it to collect massive amounts of data and metadata from pages, groups and events; and then fill in what the API dosen't do with manual data input (HTML parsing). You can quickly vacuum thousands of people's affiliations, likes, events and comments with an hour of using the API. It's easy, i've tested this before. I wrote a couple tools to do just that, and i was considering making a big NSA-tier version to put everything together, only problem is i'm a NEET who only knows Python. But i've put a lot of work and thought into this. Pic related.

2. Are you SURE you want to make it open-source? I don't mean making it commercial, i mean putting something of this caliber on the internet free for anyone to download, set up and use. Do you REALLY want the commies to pick it up and use it against us? That's the reason i've never came here and went "HELP NEEDED FOR SUPER INTERNET SPYING PROGRAM". Any work on this must be in a strictly controlled environment and total secrecy, i think we've all seen what happens when cyberweapons get into the wrong hands lately. And yes something like this is that powerful, it would be an OSINT/SOCINT data collection system the likes of which are rarely seen.

215bdb No.9928355

File: a3a41a1aa7f00c4⋯.png (174.89 KB, 304x349, 304:349, fucking bird.PNG)


reminds me of someone

ffae93 No.9928433


Your Graph API code sounds interesting. Does it allow you to get the persons profile pic or profile url? This would be an excellent source of labeled data for training the facial recognition model.

Honestly, I think the only way this will ever get finished is if it's open source. Also, I'm betting this is a side project for all of us, so it'd be nice to contribute on a come and go basis. Plus, it seems like we're sticking to publicly available information and data so it's not really a secretive spying operation. Creepy, yes. Illegal, no.

9c178c No.9928465

File: 245a61a2680dfe0⋯.jpg (259.67 KB, 778x1062, 389:531, AgtsSY.jpg)


>>9928222 (checked)


Holy cow anons! You're ding god's work. Really wishing I listened when my dad tried to teach me to code.

defb91 No.9928469


Not the profile pic, but the URL sometimes and always the user ID. The ID is a much more reliable way of identifying someone, because sometimes different people have the same name. You'd probably need in final a Person-ID different from the facebook one, because some people have alts. You could probably automate something, like a bot account, that downloads the user's profile picture. I figure you'd need a decent number of sock accounts to spread out the API requests and collection. You can buy those on blackhat SEO websites, fully activated and cellphone verified.

I know open source is the most realistic way it would ever get done, but i'm scared of putting this technology, or even the IDEA of this technology, into our enemy's heads. Amalgamation of OSINT is very powerful; it's very easy to figure out the top dogs of any movement or organization when you start analyzing. https://archive.is/KcbT6

c06b00 No.9928559


if it's not open source then it'll have to be put on a single server behind an api or web server which will eventually be shoa'd.

open source and distributed is the only way. it's not the source code that's the problem it's the database. it's likely to be filled with all kinds of shit that can be easily shoa'd via dmca. there's no way this could be done commercially for example, you couldn't charge for this, because practically the entire database would be composed of copyrighted data, either from facebook, or twitter, or youtube (the video's are copyrighted), etc.

defb91 No.9928574


Good point. So basically, put it out there and hope to god you're better than the enemy at intel and OPSEC?

Maybe you could have the database be a blockchain or something as stated above, i'm not very familiar, but it needs to be decentralized yet united. We need all the puzzle pieces to come together.

c06b00 No.9928594


the difference between /pol/ and the average leftist transfaggot is /pol/ shouldn't have their face plastered over 15 different social media accounts.

some next gen blockchain thing would be great but I doubt it's ever going to get organized, the software would just need to take multiple databases to scan over, preferably there would be some common format so these databases could be merged and duplicates removed.

the databases could then be shared over already common things, torrents, ipfs, retroshare, i2p, gnunet, w/e, wherever the infrastructure is already there.

defb91 No.9928613


The data would get pretty big tho. Hundreds of gigabytes easily once it gets serious, you would need dedicated storage for that at some point. You could just host it offshore and/or on the tor network.

c06b00 No.9928624


I've been looking at openface mostly for facial recognition, but from what it seems hundreds of gigabytes wouldn't be required for the database.

what seems to happen is it scans all these reference faces and then saves the metrics to analyze whatever your going to analyze.

you don't need to maintain the reference images afterwards, you should be able to save just the metrics (along with the other info, twitter id, etc), if you delete the images afterwards, or maybe just keep a thumbnail or something, the database size could be reduced alot.

c06b00 No.9928638


also all this facial recognition is seems to be done in grayscale, so they can be converted and saved in greyscale if you want to maintain images, which should save space

retroshare as shitty as this is could pull this off and over tor. 500gb torrents would be shitty, especially if it's getting updated on the regular. retroshare can handle the individual files and can sync and download from multiple peers etc.

c06b00 No.9928645


if the software is written to do this you could also publish massive 500gb databases semi-regular over torrents, and then bootstrap with updates, like openstreetmap does with planet files and diff's for updates, you don't download the whole planet everytime you update, you just do it once and auto-update via diffs.

defb91 No.9928659


Not just for facial recognition my man, i'm talking absolutely everything they publicly post on social media, either directly archived or linked. Of course you could disregard the irrelevant stuff, but still that's a lot of people we would be interested in, just look at my pic, >>9928318, in 2 minutes this thing spotted 4,385 users. Once you start seriously scouring through those big leftist/communist pages and groups you're looking at possibly a million people, with billions of single database entries. The power this gives you, think about it. You can learn a whole lot about someone through what he posts on the internet.


Yeah something like retroshare would seem best for the moment, you only need to get new parts and what changed.

c06b00 No.9928683


the hard part on that would be the images, way too much space would be required. if it's text data that could atleast be compressed significantly. it doesn't have to be billions of single database entries either. that's all hardcore database tweaking, but you can stick a lump of json for post data and things like that in there for each user/id.

02f49a No.9928695



my sides

ffae93 No.9928761


We need to focus small for now to just get it going. We'll adapt as we grow.

Part 1 is face detection. This is just simply finding a face/faces in a video or image.

Part 2 is facial recognition. This is comparing an "unknown" face against images of known faces and finding a match.

Part 3 is the database that stores images with other metadata like name, age, etc.

Each of these parts can initially be developed independently. The GUI is the end goal, but it needs Parts 1-3 first.

c06b00 No.9931379

File: e3caacfe5773a61⋯.jpg (91.92 KB, 905x588, 905:588, summary.jpg)


I'm working on getting openface up and running ( https://cmusatyalab. github.io/openface/ )

I'm not sure how well it handles video, but it'll definitely do Part 1 and 2. For video what it looks like it does it run it frame by frame.

c06b00 No.9931395


I don't know whats up with this "Deep Neural Network" shit either. It looks like it's just buzzwords

713bf9 No.9932669


so I should get my tech skills up?

bd38fb No.9933668

YouTube embed. Click thumbnail to play.


Anon, take a look at this video.

Does this look like a buzzword to you? Now Imagine if we a system like this active every time antifa showed up.

bd38fb No.9933682

YouTube embed. Click thumbnail to play.


And here's one with temporal tracking too.

c06b00 No.9933746



ya that's pretty cool, whatever ACT is it's clearly shit lol

c06b00 No.9933752



I take it this "Deep Neural Network" thing creates new reference frames on the fly

bdf1ff No.9933867


b00b9e No.9933977


2e65a5 No.9934283

this thread isn't the best place to actually talk about working on this, i suggest a tox group gets started

inb4 tox is shit muh ip blah blah etc, if anyone has a better suggestion by all means. tox is fine just configure it to use tor or use a vpn

bd38fb No.9934861

File: a29250c65d32e32⋯.jpg (1.5 MB, 1700x2200, 17:22, df-1.jpg)

File: 10314aac79c4ade⋯.jpg (1.48 MB, 1700x2200, 17:22, df-2.jpg)

File: ba6bded9980c324⋯.jpg (1.57 MB, 1700x2200, 17:22, df-3.jpg)


You aren't going to get a good representation from ONE face image unless you do some crazy-ass 3D shit. See pic related, that's facebook's model, and they managed to get 97% accuracy. They achieved this by replacing the detection/alignment step with a SEPARATE neural network to convert the faces to a standard 3D model, and feeding THAT into their main CNN.

To be fair though, we all know data is the real secret sauce, and god only knows The Zuck has plenty of that.


Holy shit. That's fucking impressive, thanks anon.

af3ae1 No.9935295


As long as there are good whitelist/blacklist capabilities, you'll be golden. It'll be nice to see 8chan become a faded relic.

bd38fb No.9935403

File: 4aa8f8d976dfa35⋯.jpg (806.41 KB, 1700x2200, 17:22, stylo-0.jpg)

File: 934ae0e8b09543e⋯.jpg (1.05 MB, 1700x2200, 17:22, stylo-1.jpg)

File: 8b170179ba1d18a⋯.jpg (850.43 KB, 1700x2200, 17:22, stylo-2.jpg)

File: 2db3d262f083216⋯.jpg (672.17 KB, 1700x2200, 17:22, stylo-3.jpg)

File: 79675694e3d0138⋯.jpg (662.7 KB, 1700x2200, 17:22, stylo-4.jpg)


That's not a long term solution: Computers are cheap, botfarms/botnets are cheaper, and IP addresses are free. There's no point in -keeping an ever-expanding list of banned IPs when your opponent can buy hundreds of new machines with the click of a mouse.

And If you've been keeping track of pre-existing internet subversion techniques, you'd know that the most successful schemes involve either:

1. Long-term subterfuge and narrative stearing.

2. Stylemetry over months, if not years. Pics related

bd38fb No.9935508

Why isn't anyone else commenting? This is important.

074a2a No.9935542


I still had a half written post from earlier this month.

Looking over it. I have been a dba fag for a while, trying to figure out what sort of access it would need because 500GB+ in your typical RMDBS is going to require a shit ton of ram or day long queries. There are several no-sql (most are faggy javascript), but things like Cassandra is what twitter users and it clusters nicely. I am going to be looking into that open face shit here in a bit. The closest thing I have ever done to that was to take an image and break it into an image pallet and use the pallet to search through a large file system of pictures to match thumbnails to larger pictures. It was mostly accurate, but more of a reverse image search tool.

bd38fb No.9935627

File: 4f248d07934a394⋯.png (88.4 KB, 850x644, 425:322, face.png)

File: 341251d04182e3c⋯.png (209.87 KB, 1035x515, 207:103, mtl.png)


Building a distributed database won't be the hard part, it's not a solved problem but the sheer volume of preexisting solutions means that it shouldn't be more than three weeks worth to customize one to suit our needs. I think the blockchain guys from above are already on it.

Keeping our core analysis usable is a different ballgame. OpenFace's best model is about 92.9% accurate (pic related 1), which means that it's going to be unusable after 10K identities, which sounds like a lot, but really isn't, and in practice it's going to be far less than that. Doing better is going to require a more advanced CNN solution, which is "technically doable" (pic related 2)but comes with it's own set of problems.

Case in point, Pic Related Two: FacialLandmarkDetectionbyDeepMultiTaskLearning. When I first read the paper, it astounded me, not only because of the results but because how their model/algorithm could be expanded. That being said, the maths is NOT simple, the code is in MATLAB, and while it doesn't require the best GPU, you do need one, and not everyone can run out and drop money on new hardware. The code is here: http://mmlab.ie.cuhk.edu.hk/projects/TCDCN.html. If anyone knows Matlab, CUDA, and either Theano/TensorFlow or Torch, re-implementing it into a open-source language would be amazing.

bd38fb No.9935631

File: 04ad114aa9ef892⋯.png (192.26 KB, 1434x605, 1434:605, nicola.png)


And, I just noticed that OpenFace used a TeslaK40 on there performace testing. Great.

18d326 No.9935690

Is there any way to help for someone with knowledge of a little bit of Java, JavaScript, SQL and ABAP?

bd38fb No.9935710

File: 3beccf112750654⋯.jpg (1.27 MB, 1700x2200, 17:22, diy-0.jpg)

File: b0b88f20095c0a8⋯.jpg (1.21 MB, 1700x2200, 17:22, diy-1.jpg)

File: bd0a2ce9bf56c77⋯.jpg (1.38 MB, 1700x2200, 17:22, diy-2.jpg)

File: 30c0fbdb7cde2bd⋯.jpg (1.4 MB, 1700x2200, 17:22, diy-3.jpg)

File: 829fadb37bfb6f0⋯.jpg (1.55 MB, 1700x2200, 17:22, diy-4.jpg)


Yes, read this this paper. 1/2

bd38fb No.9935716

File: a9166dcb03d64dc⋯.jpg (1.41 MB, 1700x2200, 17:22, diy-5.jpg)

File: 56787ba9ed30932⋯.jpg (1.52 MB, 1700x2200, 17:22, diy-6.jpg)

File: 7499f2c9f0be12c⋯.jpg (1.49 MB, 1700x2200, 17:22, diy-7.jpg)

File: 464f5c4bd0cef98⋯.jpg (557.66 KB, 1700x2200, 17:22, diy-8.jpg)





All of those were me, btw.

2/2. Actually, everyone should read this.

bd38fb No.9935812

And everyone's gone again.

defb91 No.9935856


Could we have a board, discord, mailing list or any other reliable way of setting this up other than a thread?

18d326 No.9935909


I set up a discord server. Join if you want, the link is going to expire in a day.


bd38fb No.9935913


I just tried to create a board, it's giving me a 403 error.

Anyway, if you manage to create one, call it /comint/ or /sigint/, or whatever you want.

af9aad No.9935960


I've tried too and I'm thinking it has something to do with April's hack. Some boards still haven't been restored yet either.

bd38fb No.9936050

File: 4bd251558495b08⋯.gif (204.87 KB, 2046x410, 1023:205, ttp2013051193t1.gif)

File: 7b968c7cbd9b189⋯.jpg (1.19 MB, 1700x2200, 17:22, occ-0.jpg)

File: 3ed3d9dae2bbc95⋯.jpg (1.63 MB, 1700x2200, 17:22, occ-7.jpg)


>Another BIG issue we have is identifying people with masks.

Actually, not really. It's called occlusion, and there's tons of research on it



Also, I can't post PDFs for some reason.

bd38fb No.9936064

Everyone who can code needs to join the discord.

9fefc2 No.9936668

File: 095f9735159e338⋯.jpg (48.58 KB, 521x800, 521:800, 095f9735159e3389fb1ec94182….jpg)


Just came by to check those trips. Carry on.

18d326 No.9938408

Here is the permanent link to join the discord


bd38fb No.9939508


Would you mind sharing your code?

Also, could you resize the images as 256x256? It would save time when using other neural nets.

4e7977 No.9939524


>The List meme

>He thinks its just a meme

bd38fb No.9939628

File: f0393f93891e177⋯.jpg (159.04 KB, 1200x776, 150:97, CmYdexbWYAASx57.jpg)

We need to come up with a unified plan or framework. Mission Creep and Feature creep have been the death of hundreds of projects.

Coming up with a single pipeline should be our first priority.

bd38fb No.9939749

Where is everyone else?

074a2a No.9939925


I will be on there in an hour or two

27fa7f No.9940083

File: 3743003573196a6⋯.png (108.32 KB, 937x742, 937:742, model.png)

Updated git; been working on the actual database layout, let me know if anyone has any suggestions.

File sharing will be done (automatically) on the blockchain; that part is actually very easy since the software already exists for it. Just gotta find the right settings and figure out what kind of file compression will be best. But, you'll also be able to import raw data into the database anyway you like; and I'm making exportable/importable filetypes for this as well so we can share them, say on 8chan or filedropper

074a2a No.9940376

confirmation post, un DIC

d364f6 No.9941074


Yeah. Pretty hilarious notion. I hope OP was trolling.

18d326 No.9941213


Can you add a field for info, which does not fit in any of these categories? Like a miscellaneous field.

9de33b No.9941644

File: c39665ff941c73c⋯.jpg (612.2 KB, 2448x3264, 3:4, oc pepe.jpg)


OpenBSD and use korn shell

nigger free with OpenBSD!

c0c561 No.9941666


Couldn't the Media-Crimes relationship be many-to-many? Especially with multiple perpetrators.

c0c561 No.9941676


Or I guess that would address multiple media of the same crime. Also HAIL

27fa7f No.9942315

File: 2033650c3886d2c⋯.png (299.14 KB, 1660x1385, 332:277, model2.png)



Yeah, I'll throw in an intermediary real quick.

An anon shared a facebook scraper with us and we're working on implementing that into the data from that into the db as well.

c166b1 No.9943014

I haven't even made it a quarter of the way through the thread yet but….


df6b98 No.9943361


What skills are required to apt-get something or click on an installer?

The problem is sustainability vis-a-vis state actors (not three letters, but simple banning/dns-fudging/filtering at ISP tier).

df6b98 No.9943386


Bash is awkward. This should be enough, shouldn't it?!

Other shells are crufty or weird, but they have structure. bash is just comical.

df6b98 No.9943410


>allow yourself to be facially recognitioned (sic!)

You're an idiot. Why would you put your own pics… what…

I did not know retards know another domain besides twatter.shitcom

df6b98 No.9943490

And here we are. Probably a simpler way to organize this dreck are simple folders which are created by shared zip/7zip files and therein are images and text files. You grep over it and some pol-dude programs a facial reco program to index local files. Look at Lexis-nexis, a glorified, multi billion dollar business of "grep over *.txt" > results-now-gimme-shekels.html

I cherish any attempt to bring back a little database clout (ha!) to the common man.

There is a bittorrent BEP for changing files. What I daydreamed once was a bittorrent style subscription sync over IP (routable through tor, i2p and cjd-ns) where a founder can pin a bucket with files and others can propose submissions. The bucket extends automatically, by pre-approved signature or block-chain style consensus algorithms.

Weird enough, what I daydreamed wasn't for pol, but a subscription to a TV series hash file where any addition would be added automatically to my torrent client and I could just let it run in the background. Long story short, I'd buy an old TFT, connect it to a Raspberry Pi and let it run 24/7. Come home in the evening, look up new stuff, go through the doxfiles, add some comments, add some data, let it spread. whynot.jpg

c0c561 No.9944333


I like the all of the additions I can see anon. It looks robust enough, I don't even think we'd have problems adding other sites, at least at a rudimentary level, but they don't organize as much as facebook.

Optional foreign key for facebook group in the Groups table?

9de33b No.9945133

File: 5a1d3dc3169458b⋯.jpg (225.97 KB, 880x580, 44:29, pepeSpitze des Morgens zu ….jpg)




stay nigger free w/ OpenBSD

best documentation and most secure os by default for a reason.

bd38fb No.9945714

Anyone who can actually code needs to join the Discord.


ffae93 No.9947615

File: 574b23d47bfa248⋯.jpg (33.4 KB, 403x348, 403:348, SWOPXeV.jpg)

Current Progress:

1) Face Detection - We have a pipeline to extract faces from videos and images. Several demos here: https://twitter.com/JohnAmericanPhD

Input = Video | Image (any dimensions, quality)

Output = Face Image (227x227 pixel, .jpg)


* Share code and details on this pipeline

* Change output size to 256x256 pxls

2) Face Recognition - There is no progress yet in this thread. There are ideas for algorithms to use, but no demos. More importantly, there should be a training dataset which can be shared among anons.


* We need images of faces with full names

* Crop face, save as "firstname_lastname.jpg" (any dimensions)

* Need a way to share this training data among anons

3) Database - We have a nice demo of the SQL schema from anon. It is currently unclear how to access and distribute this database.


* Please share SQL schema in txt form

* Need ideas for distributing database

* Need ideas for accessing database

4) GUI - Unclear about the status of the GUI. There is a demo image of "doxwizard.ui", built in Python/QT. Appears GUI function is for:

a) facial recognition of a user submitted image

b) user can submit info about a known ANTIFA member to database

NOTES : Things are looking good, we are slowly getting organized here. Biggest hurdles right now seem to be sharing code and data.

bd38fb No.9950831

File: 6a245d4462ebf24⋯.png (5.5 MB, 2000x720, 25:9, THISISYOUTHISISJERRYSHAW-0….png)

Database FacialRec Roadmap Version One

Does anyone have any questions, concerns, or ideas?

a5c5ca No.9951258


i can code anon post a 1 time invite for tox/discord/anything and we can deal with auth/verification from there

go rm xl

607f38 No.9951289

>wasn't really bothered

>until you can take images from a video and actually recognize a face

time to wear a mask

bd38fb No.9952207

File: c6b2094c5e2750b⋯.jpg (85.31 KB, 638x479, 638:479, ocll.jpg)


Not gonna help anon. Didn't you see the post above? >>9936050

a76920 No.9953524

I only briefly looked at the thread, so I have to ask: What programming languages would be best to use to program a facebook scraper to use on "Groups" of commies?

c86841 No.9953537


I'd use Python, its easy, and there are a lot of libraries available for handling javascript, json, and html parsing. Please don't try to be one of those faggots that take up something ridiculous like C++ for something like a web scraper, you'll never get shit done that way. People tend to let their e-peen get in the way of software development and go over their skill levels

27fa7f No.9954050


Python with Facepy and use the GraphAPI.


27fa7f No.9954161


>* Please share SQL schema in txt form

>* Need ideas for distributing database

>* Need ideas for accessing database

OP here, I'll take care of the database and distribution of it.

Had been working with standard relational database, but if we include facebook, linkedin, twitter, etc. with all the other data and metadata the database is going to be far too complex to be useful and extremely slow.

I've been looking into graph databases the same shit facebook and twitter use and they are 100% the way we have to go. Each account will have its' own node with a relationship pointing back to the person and vice-versa. This allows us to use many-many relationships directly instead of creating an entirely separate table; it also means we can add 'nodes' on the fly for a very flexible, changeable schema.

Here's a basic setup we can start with:





>(FirstName, LastName, MiddleName)




















Person - [is] - Person (useful for accidental duplications)

Person - [knows] - Person

Person - [lives] - address

Person - [lived] - address

Person - [has] - phone

Person - [had] - phone

Person - [member of] - Group

Person - [donor to] - Group

Person - [media] - Media

Person - [committed] - Crime

Person - [victim] - Crime

Group - [owned by] - Person

Group - [has] - Phone

Group - [address] - Address

Group - [donates] - Group

Group - [owns] - Website

Group - [media] - Media

Group - [committed] - Crime

Group - [victim] - Crime

Person <-[is]-> FBookUser

FBookUser - [posted] - FBookPost

FBookUser - [member] - FBookGroup

FBookUser - [media] - Media

FBookUser - [attending] - FBookEvent

FBookGroup - [is] - Group

FBookGroup - [follows] - Person

FBookGroup - [posted] - FBookPost

FBookGroup - [links] - Website

FBookGroup - [media] - Media

FBookGroup - [has] - Phone

FBookPost - [comment] - FBookPost

FBookEvent - [address] - Address

FBookEvent - [committed] - Crime

Crime - [at] - Address

Crime - [evidence] - Media

Crime - [evidence] - Website

Crime - [suspect] - Person

Crime - [suspects] - Group

Crime - [happened-at] - FBookEvent


As far as the database is concerned; it can be adapted and changed easily to allow data from all major social media outlets.

In fact we can use FBook* for pretty much every other social media outlet as well and simply add a new relationship e.g.:

Person - [twitter] - FBookUser

Person - [blog.wordpress.com] - FBookUser

[Return][Go to top][Catalog][Post a Reply]
[ / / / / / / / / / ] [ dir / cute / fur / htg / lovelive / misr / wai / waifuist / wooo ]