Show HN: A Hand-Drawn QR Code Alternative • Hacker News Board

Re: Show HN: A Hand-Drawn QR Code Alternative

#121

Post by IanCal » Tue, Jul 28, 2015, 5:43 AM UTC

sherjilozair wrote: post #93

CPLX wrote: post #7

It's a confusing solution for a problem that does not exist, which is how to use a pen to draw readable shapes, a problem many of us learned how to solve in elementary school.[0] [0] http://www.auburn.edu/academic/education/reading_genie/lette...

Sorry, but the problem is how to use a pen to draw machine readable shapes. Machines find reading QR codes and OP's triangles easier than reading letters.

If you wrote numbers or letters as precisely as the triangles are drawn in the example, I doubt computers would have much of a problem reading them. We've got fast algorithms for digit classification that are pretty much perfect on the MNIST dataset with just 0.21% error (and I'm pretty sure there are errors in the dataset itself, which brings the real error rate down). MNIST is an easy image recognition task compared to others but it's not done with clearly written numbers.

Here's an example of some that were misclassified by an algorithm: http://www.concordia.ca/content/concordia/en/research/cenpar...

Re: Show HN: A Hand-Drawn QR Code Alternative

#122

Post by amelius » Tue, Jul 28, 2015, 7:05 AM UTC

How many bits can it encode? I bet not many. And I bet it is far easier to just write down two or three characters like "Am3" or "IX8" and use OCR to extract the information.

Re: Show HN: A Hand-Drawn QR Code Alternative

#123

Post by jarofgreen » Tue, Jul 28, 2015, 8:19 AM UTC

bengali3 wrote: post #56

Sam, Cool stuff. I get what you're doing here, but from these comments I think QR code reference really throws people off. While comparing to QR might explain how it works, I would do my best to position this as nothing like QR codes (and highlight use cases that make it unique from QR) to avoid the comparison.

This. All the concerns first parent raised (SPOF, etc) are valid ones, but that doesn't mean the service is inherently bad, and I'd hate Sam to take all this as a completely negative response. It just suits a different use. I get that it's aimed at people who want to hand draw the symbols, and that's pretty cool.

Re: Show HN: A Hand-Drawn QR Code Alternative

#124

Post by jarofgreen » Tue, Jul 28, 2015, 8:20 AM UTC

samfpetersen wrote: post #113

davidbanham wrote: post #105

Earlier quoted context omitted.

I'm less worried about the requirement for an internet connection than I am about the single point of failure. When the centralised service goes down, all your tags are useless squiggles.

It's a pretty simple back-end - cheap and easy to run. I don't plan to take it down!

Sure, but you have to admit that history is littered with things like this where they still died anyway. Can you not see how people could have concerns over this?

(I've commented elsewhere on this page BTW; I don't think the fact these are valid concerns makes it inherently bad ...)

Re: Show HN: A Hand-Drawn QR Code Alternative

#125

Post by leni536 » Tue, Jul 28, 2015, 9:01 AM UTC

I would only be interested in the retrieval of the raw data of a meshtag. It's of course smaller than the shortened url. I always wanted to make a simple method to mark my handwritten notes, scan them, then the scanned pages "organize themselves" by the hand drawn tag (in this case meshtag) at the top right corner.

Re: Show HN: A Hand-Drawn QR Code Alternative

#126

Post by leni536 » Tue, Jul 28, 2015, 9:15 AM UTC

pnut wrote: post #66

I mean this honestly. Do people actually even use QR codes? Filling up your phone with a bunch of hi-res pictures of bar codes, not to mention having to load: 1. QR code app 2. camera app 3. browser app These things are 100% in the gimmick column in my life.

In Hungary the train company sells home printable tickets and they use QR code to ID it. I think it's weirdly encoded, maybe even encrypted somehow. You don't have to print it though, you can just open it in your phone than the ticket control scans it with their own phone and in-house app.

As for personal use, I don't really know. Storing your larger bitcoin wallet on paper could be reasonable.

Re: Show HN: A Hand-Drawn QR Code Alternative

#127

Post by camillomiller » Tue, Jul 28, 2015, 9:54 AM UTC

mrb wrote: post #76

This is nothing more than a (pointlessly!) complex URL shortener... Meshtags are exactly equivalent to the following hypothetical service which I call "SuperNumericShortener": 1. Pick any random number, say 1234. This is your identifier. 2. Register this identifier and your URL at SuperNumericShortener.com 3. Write "SNS-1234" anywhere you want. 4. Install a fancy smartphone app that can take a picture of a writing an…

What a great idea!!!! Thanks! I'll send you a check when I'm rich.

Re: Show HN: A Hand-Drawn QR Code Alternative

#128

Post by camillomiller » Tue, Jul 28, 2015, 9:55 AM UTC

IanCal wrote: post #121

sherjilozair wrote: post #93

Earlier quoted context omitted.

Sorry, but the problem is how to use a pen to draw machine readable shapes. Machines find reading QR codes and OP's triangles easier than reading letters.

If you wrote numbers or letters as precisely as the triangles are drawn in the example, I doubt computers would have much of a problem reading them. We've got fast algorithms for digit classification that are pretty much perfect on the MNIST dataset with just 0.21% error (and I'm pretty sure there are errors in the dataset itself, which brings the real error rate down). MNIST is an easy image recognition task compare…

Yep, but drawing readable triangles is much easier for a lot of people

Re: Show HN: A Hand-Drawn QR Code Alternative

#129

Post by ricardobeat » Tue, Jul 28, 2015, 11:12 AM UTC

IanCal wrote: post #121

sherjilozair wrote: post #93

Earlier quoted context omitted.

Sorry, but the problem is how to use a pen to draw machine readable shapes. Machines find reading QR codes and OP's triangles easier than reading letters.

If you wrote numbers or letters as precisely as the triangles are drawn in the example, I doubt computers would have much of a problem reading them. We've got fast algorithms for digit classification that are pretty much perfect on the MNIST dataset with just 0.21% error (and I'm pretty sure there are errors in the dataset itself, which brings the real error rate down). MNIST is an easy image recognition task compare…

Machine-readable codes also have space-efficient error correction built in, something that plain text does not support. A text label can become unusable with the loss of a single character.

Re: Show HN: A Hand-Drawn QR Code Alternative

#130

Post by dibbo » Tue, Jul 28, 2015, 11:21 AM UTC

pulkitanand wrote: post #99

dibbo wrote: post #51

hey, this is wonderful. it would be great if street painters adopted this to communicate with their audience. i would love to go into a tunnels with cool street paintings and scan these signs and get access to playlists, videos and messages that the painter wants us to visit. these cool triangles could become symbols of an underground movement if promoted in these communities.

Except that the meshtags won't last forever by design.

"Small meshtags have lifespans that depend on their complexity. A simple tag might only last for a week, while a more complex one might last for 6 months or a year. Very complex tags will never expire. This prevents the simple tags from being used up." So the mantra is, i guess, make it complex. Also leave it to the imagination of a street painter, i am sure they will make art out of it.