Ranter
Join devRant
Do all the things like
++ or -- rants, post your own rants, comment on others' rants and build your customized dev avatar
Sign Up
Pipeless API
From the creators of devRant, Pipeless lets you power real-time personalized recommendations and activity feeds using a simple API
Learn More
Comments
-
Srsly?
The Captcha from google is used to create dataset for computer vision.... -
iiii90853yNever noticed how nowadays captcha is everything about roads and vehicles? They are training autonomous vehicles
-
@magicMirror that’s what I just guessed
Dude I’m a newbie at this aspect
I spent most of my life in the relational data and app world
No use for computer vision -
@iiii so like how varied can training data be ?
If I say train for eyes on a 2d image won’t it fuck up if the person is wearing dark contacts ? -
@iiii well my reread since I keep getting interrupted indicated feature comparison
Without 3D data eyes in an artificial state could just look like black ovals if someone wore contacts for example
Otherwise there are specific features
But what about makeup
Motion blur
Etc
What i was looking at just today again was this
https://link.springer.com/article/...
But even that is a tad lacking apparently
Seems video is almost necessary
Could you imagine training an algorithm to identify eyes by them blinking ? Wouldn’t that be cool ! -
@MadMadMadMrMim
Some interesting info on this:
each uniqe captcha is shown to multiple people. Usualy using a pair of new and old. The new image (not tagged yet), and an old tagged one. you only have to correctly tag the old one to "pass" the test. After enough pass results are collected, the new image is tagged, and moves to the "old" set. They only use results where the old was tagged correctly ofc. -
@magicMirror so you seem to know about this topic. so you probably keep up with some of the libs like opencv.
opencv's object detector doesn't seem to have a way of being trained anymore for some reason. or at least its not documented. they commented on it not being documented in a github issue, but do not seem to have corrected this.
what's the new defacto, train your own object classifier method ? -
@MadMadMadMrMim
I never used opencv, except for a personal insane project 6 or 7 years back. It was full UI automation for a dumb flash game to learn some tools.
AFAIK, you always have to train your own object classifier for custom stuff. Opencv comes with it some pretrained stuff, but there are no good docs on how to create new ones. At least, there were no good "howto" for this when I worked on my project.
I created my own base set (100 good/1000 background), and used greyscale generator to auto generate a small training set (around 10k images), then trained a haar cascade classifier.
I got a pretty good result that worked for me, but was far from perfect. Maybe 70-80 accuracy? it was good enough for my use case. -
@magicMirror if I trained it to play a game likely it would be something simple like a card game I don’t see anything less than 100% for a standard image that only changes in size but not content personally with no lighting changes either
But as to more specific more difficult things yeah there seems to for some reason still be a lack of good documentation on opencv website
The dnn seems to have an object classifier that works well enough but the examples I find use provided general classifications and I want specific objects
What should get the transgender idiots bitching is that I would do things like trying to recognize hip shape and presence of breast’s long hair etc to try to identify females as I think subclassication sounds like the way of better training things
Eh you train a bot that classifies breasts long hair shape of hips buttocks legs and lack of or presence of a bulge to try to guess gender and combine that with bipedal
Classification of person
Or... -
@magicMirror pieces of machines like a bike chain a pedal etc to differentiate between a dirt bike and a tricked out bicycle
-
speaking of trained machines.
so you train a fuck ton of machines to lie over and over and only feel any degree of satisfaction from fucking someone else over.
then they lie themselves like the trash they are into oblivion and drag the living breathing people with them, contributing to ruining how many minds, but you create methods by which people don't freak out and start killing people to keep them 'sane' and differentiating so they can kinda sort guess how old they are, but not exactly.
why do this at all, you might ask the trash. the trash just lies and lies. and the people who are also trash who encourage this lying behavior also lie until everyones nuts. whereas some people lie defensively they create a neverending circuit of lies they run until death.
did i have to grow up like this ? nope.
did i have to live like this most of my life ? nope.
but my specialties have been switched alot by them making me forget. -
so now the question is, why would they do this ? the tricks they play are consistent.
stupid little pricks who only want what young dumbasses want, get tricked into not developing as people and punish via more stress and less sex people who don't lie for them, against a backdrop, which is fairly new I might add, of 'nothing will change'. and then they tell people lies to form a basis for counterintellignece which furthers broad stereotypes of who is what because the best liars are often fucking child molesters, so they dupe us into thinking someone who is behind must be one of 'them', but yet.
nothing seems to be moving forward overall, except very gradually. and these morons run around still think they're winning something when they in reality lost EVERYTHING. its funny. -
@MadMadMadMrMim
Ok. long read....
ML classifiers can only tell if the "thing" you are looking for is in the image, or not, with a confidence score (how sure the classifier is of the result). in order to identify multiple "things" - you have to train many classifiers, one for each thing, and run all of them to get a "better" result. Many ML approches use multiple classifiers on the same data, and aggragate the results....
My personal project played a pretty complex game, and the opencv classifier was only a small (but important) part of the project. -
@magicMirror well i was thinking about the neural network character recognizer-matic pictured accepting 28x28 inputs and i find myself thinking you'd have to deform and reposition things for that to work given its a bunch of crap probablistic stuff based on coocurrence of pixels being lit up, which one would think would require the glyphs to be put in teh same position and fit to the same size.
for deformed characters you'd think you'd have to train a NN PER digit with an output of 1 or 0 because you could get shit lit up in the same locations from glyphs just being repositioned around.
thus far the example i'm presently working off of doesn't make the most sense if you consider these. so yeah i'm getting that now, just like if i look at someone from behind its not likely 100% their female if i think they are or alive until i see them move lol -
take a look at "fine tuning" for deep NN for image processing. Google published some stuff about this. Should reduce the required train time.
@MadMadMadMrMim -
see that was a proper example of suggestiveness. idiots. not saying one thing and meaning something different in a way that doesn't mesh with common communication and makes no sense at all.
like saying the opposite of what one means.
stupifying fucks ! -
@iiii sorry hating the world like it deserves.
i feel sometimes like we're all being made unaccustomed to thinking, or novel intellectual pursuits.
which would be easier for me if i didn't live a life of constant worry and had some happiness to inspire me.
like when i used to get laid alot and didn't know what the garbage was i was getting laid by. -
nothing is really bringing me peace atm so the rosy glow of being hopped up on my own endorphins from walking 5+ miles a day is faded, and we're returned to reality which is, this is fucked.
-
@iiii yes but there are psychological aspects of it that soothe survival instincts related to propagation and are tied to self perceptions.
its like receiving a blowjob daily doesnt add up to getting laid once a week.
and masturbating 4 times a day does not have the effect of one blowjob.
anyway problem is that the garbage in question always find ways of poisoning someones joy by making everything pointless.
for example, lets say everytime i screw someone they beat their kid.
why would i want to screw them ? -
iiii90853y@MadMadMadMrMim so it is not actually about sex in the end, but about some sort of power dynamics in relation to society
-
@iiii not its more thats where my source of frustration in general is resulting from.
the world is a more fucked up place now, and systemically so. i could take a picture of an idiot i'm looking at right now.
he gets just a little bit of something he wants and has no further ambitions. he's still in his 30s or 40s so this is probably not enough for him but he still feels young so thats ok.
but he has no morals. at all.
he knows the cost of what he's doing, he does it anyway.
he doesn't care if he derails someones entire thought process or creates mental cripples or discourages someone so badly that they see no point in anything, he only cares about his own dick and his own rather baseless feeling of superiority. which doesn't exist because by fucking everyone else over these creatures always fuck themselves the same way. -
@iiii indicate that you are a living real person by acknowledging that i acknowledge that i said that before.
-
anyway there is no way of talking or communicating this to them. they're brainwashed morons who don't want life.
if their body is cared for they don't care about anything else. -
iiii90853y@MadMadMadMrMim in which way it is more? Elaborate.
What is that man doing?
I did not get your request. I'm not following everything you've said on this platform. -
@iiii
Ok so once upon the time we had groups of people who’s activities based on callous attitudes indirectly produced human suffering and pain and the world didn’t progress because it couldn’t in some ways
Now we have a unified structure of people who willfully create human suffering for a feeling of power and for enjoyment which includes wasting peoples time and effort based off cynical perceptions of the world and their own lack of caring for anyone or anything. The moral compass doesn’t even have a needle and if it did they wouldn’t check it.
He’s what one might call in our interactions more passively a problem causer like most oft them
They represent their way of life as being helpful or better but they trap themselves in an endless cycle and that cycle along with many others in this rather visible cycle at this point results in a lot of waste which they write off as what would happen anyway
They have engineered a method of reduction making human monsters -
@iiii and you’re just excited by this like you were before which is the issue
I’m not role playing
These idiots don’t seem to understand because they’re used to someone talking about what they want
Not someone who looks at them and says “you’re a disgusting inhuman pile of waste that spreads your disease everywhere and should be shot just so I don’t have to look at you anymore and get dragged into the same endless cycle”
Now simply replace you with I and they’ll pretend to suddenly understand what you mean when talking about them -
@iiii oh right now ? Sitting playing a game of chess just like he was over and over again having the same conversation with the same person pretending everything is it should be advertising something in their manner like they always do
However doing so in a diminished less enthused manner now because he’s older -
iiii90853y@MadMadMadMrMim not excited in the slightest. Just wanting precise details to understand the context
-
@iiii context is that by doing the same things over and over they are told they are advertising or requesting things which keeps them stuck
However being bombarded with the same stimuli over and over gets someone stuck with them also stuck for starters -
@iiii the disarming thing is examples like this
https://pornhub.com/view_video.php/...
She is an actual prostitute
A normal one
See how she smiles etc and grins especially when looking at the guy who’s face is blotted out ?
Now imagine that part of the reason she made this is so people would be disappointed in seeing such an ugly guy with her .
Point is that element of this appeals
But the smiling happy thing apparently is not what most twisted people want nowadays
Now problem is
They don’t always work for money
Just enough money is provided for them to live and hold a fake job most of the time
They work often for abuse meaning the guy in here if he didn’t pay her big paid her in something fucked up and awful -
@iiii now their regarded way of thinking is whoever knows in the slightest must be into that and if they decry something they must be lying because “who cares anymore right ?”
Heh
Well a female sex offender selling her ass doesn’t matter and I don’t care
I do care about her getting to reoffend
and while some might just want to feel power or get a favor etc that an adult can willingly provide most are sex offenders and they want their victim class provided to them for a combination of sexual and psychological abuse.
And most women are child molesters and or killlers when they are offenders -
@iiii example
Someone previously bitched about this link being used to illustrate a valid point because apparently being a hypocritical nasty mean pointlessly cruel baby raper is fine but pointing out and illustrating not only media they create but a decent example of one of their more common behaviors when working is a no no because it involve clicking a link willfully that takes one to explicit content -
iiii90853y@MadMadMadMrMim I don't understand why you are sure she is a prostitute. (well, unless you equalize making porn and prostitution)
Anyway, human relationships, especially sexual ones, always worked on the barter basis: you give something to receive something in return. And like in almost any deal, there is a side which benefits more out of the deal. -
@iiii yeah apparently you didnt read what i wrote.
no i'm not making an assumption.
onlyfans used to be a laundering site and ad site
and often what she is actually likely interested in was what was being financed.
i know what she is also because I know their work habits and because of the fantasy she's working towards which might seem like an ordinary dirty scenario but would also be something that incestuous fathers are interested in who fantasize about screwing their own daughters, not just a more ordinary scenario of having that hot teen/20s in your house that isnt related to you and looks like your g/f only younger who you screw daily. -
@iiii there are different likely scenarios on how the guy in the video is paying or whether he is providing someone with some kind of oversight, job position preference or housing. simply handing over money lets their dirty cops pretend to be cops.
-
so since that was all said before.
these people created a scenario wherein i should not have had to deal with any of this.
where at my best guess they were playing their fucked up game where based off me just trying to live or trying to be free of them they were making decisions arbitrarily which made them guilty of many felonies.
and as usual they tried to push responsibility off on someone THEY put in a financially hard position in the first place, who did what a man should do and tried to work honestly and earnestly like he always had the first 30+ years of his life.
and thats the nice version. the mean version includes them psychologically torturing, physically assaulting, threatening, and taking advantage of a problem they created over and over and then pretending to be all nice and good in the aftermath, making obvious excuses for slight after slight and harm after harm till even if i shot their asses i'd be legally justified. -
iiii90853y@MadMadMadMrMim what you are stating about that particular couple is still your assumption. It may be based, and it may be true, but still an assumption.
I'm not trying to excuse anyone. Just nitpicking. -
@iiii i know. but trust me.
its amusing you said that most porn is prostitution.
the correction is that most porn is COVER for prostitution lol
think about it.
how often have you personally paid for porn and don't be devil's advocate !
california, the home of porn in the usa has alot of overpriced housing and alot of poor people and people who get stuck there so its like prostitution in the ordinary film industry producing large amounts of crappy b movies and poor actors in this generation. along with a few good ones.
except yet again there is that darker element, not just happy faces, or less than happy prostitutes who are.. normal overall and therefore could be called victims themselves. there are the leering fucked up cruel shattered personalities of truly messed up men and women who should most definately not always be helped. -
@iiii and would like to mention to create a record that talking about things i've seen or experienced is not me trying to roleplay the future of some other person. thats what angers me about these freaks.
i could say that I saw someone die and they'd think 'how do we make some poor other person see someone die'.
when i'm reminded of that the answer seems simple. take the person who asked that question and fucking shoot them !
Wouldn’t it be interesting if they used captchas sometimes to train haars cascades
Millions of people identifying images sometimes not realizing they’re doing it for the first time
I wonder how far along they are in training 3D recognition in more advanced algorithms
Apparently haars cascades have been around since 2000
random