My mom died when I was 7, after which my dad bought me a Commodore 64 so I had something to lose myself in during the mourning process.

I learned everything about that system, from my first GOTO statement to sprite buffers, to soldering my own EPROM cartridges. My dad didn't deal with the loss so well, and became a missing person 5 years later when I was 12.

I got into foster care with a bunch of strict religious cultists who wouldn't allow electronics in the house.

So I ran away at 14, sub-rented a closet in a student apartment using my orphan benefits and bought a secondhand IBM computer. I spent about 16 hours a day learning about BSD and Linux, C, C++, Fortran, ADA, Haskell, Livescript and even more awful things like Visual Basic, ASP, Windows NT, and Active Directory.

I faked my ID (back then it was just a laminated sheet of paper), and got a job at 15-pretending-to-be-17 at one of the first ISPs in my country. I wrote the firmware and admin panel for their router, full of shitty CGI-bin ASP code and vulnerabilities.

That somehow got me into a job at Microsoft, building the MS Office language pack for my country, and as an official "conflict resolver" for their shitty version control system. Yes, they had fulltime people employed just to resolve VCS conflicts.

After that I worked at Arianespace (X-ray NDT, visualizing/tagging dicom scans, image recognition of faulty propellant tank welds), and after that I switched to biotech, first phytogenetics, then immunology, then pharmacokynetics.

In between I have grown & synthesized and sold large quantities of recreational drugs, taken care of some big felines, got a pilot license, taught IT at an elementary school, renovated a house, and procreated.

A lot of it was to prove myself to the world -- prove that a nearly-broke-orphan-high-school-dropout could succeed at life.

But hey, now I work for a "startup", so I guess I failed after all.

AI developers be like

Not so much screaming as staring in disbelief, mumbling profanity in his direction...

When my department lead said "I don't think this unit testing hype or code reviews make much sense, it's more efficient to just make a checklist and test the application yourself"

This was the QA department of an aerospace company, we wrote NDT software to do image recognition on xrays of alloy welds and micrometer laser measurements on fuel tank surfaces. Software which is quite mission critical, a single misrecognized welding fault could literally cost up to half a billion dollars — not to mention that it's a very sabotage & espionage sensitive industry.

After raising some hell he was replaced though.

The best motivation to quit Image Classification... x_x

!rant

Just wanted to share stuff. It's my first time.

<backstory>
I'm a c# dev, recently got excited about neural networks and stuff. I have a gf who studies biology
</backstory>

So i've noticed yesterday what my gf is doing for her science stuff. She has an image taken through a microscope of some erytrocytes and shit. And she's clicking on those tiny fuckers to count them. There are like almost a hundred of those things in an image and she has a butload of those images.

I was like "what the fuck? Don't you have an app that counts the stuff for you or something?"

And there is none. Or at least i wasn't able to find one. That's bullshit. My inner programmer screams with hate for boring repetitive tasks.

So i guess i'm going to write a neural network to count similar stuff in an image.

Was a co-lead on the first project to add image recognition to ATMs for depositing checks. Turns out, it was a pretty OK idea and people liked it.

Google: "How do we have millions of people identify images for us so we can reinforce our image recognition software?"

Answer: reCAPTCHA

Just to clarify thing, FaceID isn't the same tech as what we've had on Android.

In Android, it's based on image recognition. That's the reason it was so easy to bypass with a high resolution photograph.

In FaceID, it projects thousands of dots on your face and creates a depth inclusive map which is used for verification. That's the reason why it's supposed to work even if you have glasses on, etc

So please let's stop with the comparison

Just now I realized that for some reason I can't mount SMB shares to E: and H: anymore.. why, you might ask? I have no idea. And troubleshooting Windows.. oh boy, if only it was as simple as it is on Linux!!

So, bimonthly reinstall I guess? Because long live good quality software that lasts. In a post-meritocracy age, I guess that software quality is a thing of the past. At least there's an option to reset now, so that I don't have to keep a USB stick around to store an installation image for this crap.

And yes Windows fanbois, I fucking know that you don't have this issue and that therefore it doesn't exist as far as you're concerned. Obviously it's user error and crappy hardware, like it always is.

And yes Linux fanbois, I know that I should install Linux on it. If it's that important to you, go ahead and install it! I'll give you network access to the machine and you can do whatever you want to make it run Linux. But you can take my word on this - I've tried everything I could (including every other distro, custom kernels, customized installer images, ..), and it doesn't want to boot any Linux distribution, no matter what. And no I'm not disposing of or selling this machine either.

Bottom line I guess is this: the OS is made for a user that's just got a C: drive, doesn't rely on stuff on network drives, has one display rather than 2 (proper HDMI monitor recognition? What's that?), and God forbid that they have more than 26 drives. I mean sure in the age of DOS and its predecessor CP/M, sure nobody would use more than 26 drives. Network shares weren't even a thing back then. And yes it's possible to do volume mounts, but it's unwieldy. So one monitor, 1 or 2 local drives, and let's make them just use Facebook a little bit and have them power off the machine every time they're done using it. Because keeping the machine stable for more than a few days? Why on Earth would you possibly want to do that?!!

Microsoft Windows. The OS built for average users but God forbid you depart from the standard road of average user usage. Do anything advanced, either you can't do it at all, you can do it but it's extremely unintuitive and good luck finding manuals for it, or you can do it but Windows will behave weirdly. Because why not!!!

Maybe it's old and well known, but somebody asked, so here it goes:

A shepherd is quietly grazing his sheep on the fresh village pastures.

Suddenly a shiny new car stops by. A cool guy, very well dressed hops out and asks him: "Good man, If I guess exactly how many sheep you have, can I win one?"

The shepherd, puzzled, accepts.

The cool guy, opens his laptop, download a satellite picture of the area, run a NASA algorithm for image recognition and in few seconds answers "you have 1342 sheep"

"Wooow" says the shepherd "you won, take one"

The cool guy is about to live when the shepherd approaches him:

"Ehi, Young man, I bet all my flock against your car that I can guess what is your job"

The cool guy, (he likes to bet after all) accepts.

"You are a consultant" says the shepherd.
"WTF! how did you even..."

"Well, easy" says the shepherd "you came out of nowhere, well dressed and smart looking, you answered a question nobody asked you, you told me something I already knew, you want to be payed for that and in addition, you don't understand shit about my business."

"Now", adds quietly the shepherd "please, give me back my dog"

(for @LOLjustCoding)

Regarding Article 13 (or 17 or wherever it moved to now)… Let's say that the UK politicians decide to be dicks and approve the law. After that, we need to get it engineered in, right? Let's talk a bit about how.. well, I'd maybe go over it. Been thinking about it a bit in the shower earlier, so.. yeah.

So, fancy image recognition or text recognition from articles scattered all over the internet, I think we can all agree.. that's infeasible. Even more so, during this lobby with GitHub and OpenForum Europe, guy from GitHub actually made a very valid point. Now for starters, copyright infringement isn't an issue on the platform GitHub that pretty much breathes collaboration. But in the case of I-Boot for example, that thing from Apple that got leaked earlier. If that would get preemptively blocked.. well there's no public source code for it to get compared against to begin with, right? So it's not just "scattered all over the internet, good luck crawling it", it's nowhere to be found *at all*.

So content filtering.. yeah. Nope, ain't gonna happen. Keep trying with that, EU politicians.

But let's say that I am a content creator who hates the cancer of joke/meme because more often than not it manifests itself as a clone of r/programmerhumor.. someone decides to freeboot my content. So I go out, look for it, find it. Facebook and the likes, make it easier to find it in the first place, you dicks. It's extremely hard to find your content there.

So Facebook implements a way to find that content a bit easier maybe. Me being the content creator finds it.. oh blimey! It can't be.. it's the king of freebooting on Facebook, SoFlo! Who would've thought?! So at that point.. I'd like to get it removed of course. Report it as copyright infringement? Of course. Again Facebook you dicks, don't make it so tedious to fill in that bloody report. And look into it quickly! The videos those SoFlo dicks post is only relevant in the first 48h or so. That's where they make the most money. So act more quickly.

So the report is filled, video's taken down.. what else? Maybe temporarily make them unable to post as a bit of a punishment so that they won't do it again? And put in a limit to the amount of reports they can receive. Finally, maybe reroute the revenue stream to the original content creator instead. That way stolen content suddenly becomes free exposure! Awesome!

*suddenly realizes that I've been talking about the YouTube copyright strike system all along*

… Well.. maybe something like that then? That shouldn't be too hard to implement, and on YouTube at least it seems to be quite effective. Just imagine SoFlo and the likes that are repeat offenders, every 3 posts they get their account and page shut down. Good luck growing an audience that way. And good luck making new accounts all the time to start with.. account verification technology is pretty good these days. Speaking of experience here, tried bypassing Facebook's signup hoops a fair bit and learned a bit about some of the things they have red flags on, hehe.

But yeah, something like that maybe for social media in general. And.. let's face it, the biggest one that would get hurt by something like this would be Facebook. And personally I think it's about time for that bastard company to get a couple of blows already.

What are your thoughts on this?

Tech head fires a mail few days company is planning a hackathon. Overnight at office, with food, music and home drop offs in morning. We devs feel excited we will get time to work on our personal projects and complete them.

Yesterday, tech head fires a mail about the topics. Guess what? The topics are projects which company needs to scale up... Image recognition and text extraction. Selenium. Esign.

Now I am searching for an excuse to skip the hackathon...

So easy to make typographic attacks on image recognition models.

Depending on your implementation, you may need to change your entire model.
FML.

The struggle with installing opencv is fucking real!
FUCK YOU PIECE OF A SHIT

Fuck accessibility.

No WAIT, before you call me an asshole hear me out.

So when you use CSS grid to create layouts you're supposed to not use the features it has (reordering items) too much, and instead keep the HTML structured the way it's supposed to be read.
When you add a picture of a cat you're supposed to put a alt="Brown cat sitting on a chair" there.
Also you should test for all kinds of sight disabilities and use high contrast colors.

All that for likely <1% of your users.

What would be the alternative? HTML is a markup language, and not supposed to be directly read by humans. Invest the time ONCE for screenreaders to understand CSS positioning and read content in a sensible order. Use image recognition to describe pictures (with selectable levels of detail). Let the browser modify colors on the fly for better readability.

Don't spend time and money to solve a problem 100000 Times that could be solved once.

Fuck accessibility.

Given a competitive multiplayer simulator game where you can program your own flight computer, what programming model would you like it to support?

- in-game programming with a DSL which can have artificial resource limits to regulate the extent of automation and can maintain the atmosphere of the game in the process of programming

- in-process Webassembly executor which makes artificial resource limits feasible to the extent that people can't just throw image recognition and AI at every problem, but it loses the atmosphere

- API-based unrestricted remote control

# This isn't THAT bad, but since I never had any collab before this one, this is the worst so far

I'm in a web development school where we need to do a yearly project. At the beginning, we started with the idea of doing an online wallet that would handle crypto-currencied (#blockchains), and other currencies too.
On the paper that sounds good, but the dude decided to create a NodeJS server api, and let's be honest, this was a gas factory. I couldn't help him because he was too fast in his ideas, and the third member was a bit more useful because he was the one creating the mobile app, so all he needed was an url that the dude couln't manage to create.

After a few weeks he started over the project, then over again a few weeks later, before coming to us and saying it was too difficult. We said "yeah, I mean you're own your own since the beginning, no wonders!" "Uh do you guys care if we change the whole project to do something else? Like a CV library"

Went a moment where he tried to over sell some incredible (read "overly common") features that already existed 10 years ago on some famous websites (ie. Monster), and he then eventually told me that this idea came from his new job, and that they needed this library. So we would have to work for his company for free. Nice.
The third guy and me came with a new idea (image recognition with IA and stuff), and we saw the dude maybe 5 times the whole week while we're supposed to work together

So I am opencv curious. I started wondering how to train a recognition model for a particular part of the body. Everything I search wants to talk about body recognition and face recognition. I did find one article that talks about training say for hands or feet or body posture. So to satisfy my curiosity I am going to be playing with opencv this weekend. I think it will be fun!

Any suggestions for a good starting point for learning to do more with neural nets? Not interested in image recog so much, but would like to see the cutting edge of textual pattern recognition... I dunno, I don't even want my expectations to color this... whats do you guys find most interesting and enjoy playing with? Python is preferred but I'm grateful for any tips/links/ideas/rants you might share!

Obviously ai and autodocument recognition and data extraction is not usable yet
Excepting when it's a pdf not a scanned document or image
Ocr may be but shift the whole.image or bend it or remove a border from some white out
And then handwritten

ML and NLP people I need some help bc I’m stupid

I want to build a model that is trained on a set of images and captions for those specific image.

The images are of a single person in different positions and different environments.

How do I train something like this? Do I use an object recognition model to understand what’s in the image as it’s features?

Cereal is one of the most popular breakfast foods worldwide. According to the National Cereal Day website, Americans consume approximately 100 billion bowls of cereal every year. With such a high demand for cereal, manufacturers are constantly looking for ways to make their products stand out. Custom cereal boxes are an excellent way to differentiate your product from competitors. In this article, we will explore the benefits of custom cereal boxes.

Enhanced Brand Recognition

Custom cereal boxes allow your brand to stand out on the shelves. With unique packaging, your product is more likely to catch the attention of potential customers. When a consumer sees a distinctive box design, it becomes easier for them to remember your brand, making it more likely they'll purchase your product in the future. Custom packaging can also reinforce your brand's messaging and values. Whether you want to promote a new product line, a charitable initiative, or just your brand's logo, custom cereal boxes provide an excellent opportunity to showcase your brand in a visually appealing way.

Increased Product Appeal

Custom cereal boxes can make your product more appealing to consumers. Unique designs, vibrant colors, and creative patterns can create an emotional connection with your target audience. Your custom cereal boxes can be used to convey the quality of your product, the nutritional benefits, and the flavor. Consumers will be more likely to pick up your product and try it out if they are attracted to the packaging.

Competitive Advantage

The cereal market is highly competitive, and custom cereal boxes can give your brand an edge over competitors. With unique packaging, your product will stand out among the other cereal boxes on the shelves. Custom cereal boxes can also be used to create a sense of exclusivity, making your product more desirable to consumers. Consumers are more likely to purchase a product that appears to be of higher quality, and custom cereal boxes can help create that perception.

Improved Customer Experience

Custom cereal boxes can improve the customer experience by creating a memorable and enjoyable shopping experience. Unique packaging can create a sense of excitement and anticipation for the consumer. Additionally, custom cereal boxes can provide useful information to the consumer, such as nutritional facts, ingredients, and serving suggestions. Consumers are more likely to have a positive experience with your brand if they feel informed and engaged.

Eco-Friendly Options

Custom cereal boxes can be designed with eco-friendly materials, making them an excellent choice for environmentally conscious consumers. With the rise of eco-friendly products, custom cereal boxes can help your brand appeal to consumers who are looking to make more sustainable choices. Additionally, eco-friendly packaging can reduce waste and promote a positive image for your brand.

In conclusion, custom cereal boxes offer many benefits to cereal manufacturers. From enhanced brand recognition to improved customer experiences, custom packaging can help your brand stand out in a crowded market. Whether you're looking to promote a new product line or create a sense of exclusivity, custom cereal boxes are an excellent tool to help you achieve your marketing goals.
If you are also looking to increase your sales, get custom cereal boxes from OXO Packaging.