Hacker News Reader

Comment by yantrams | Hacker News Reader

Author here. Happy to answer any questions, hear feedback.

I spent some time confused by the animated running horse and how it was supposed to be related with the other icons. Until I found out that it was the loading gif.

I would change there for something more standard.

I found the running horse to be an adoring touch. The author's site is also worth checking out: http://linkdot.link

https://en.wikipedia.org/wiki/Sallie_Gardner_at_a_Gallop

I believe if it would be both adorable and more easily understandable if it was a smaller image, giving a visual hint that is fulfills the role of an icon, not a featured image

It was the other way for me, I linked it to the duckduckgo icon svg and the horse started. I thought it was a loading animation (and took more than a few seconds) so I threw it to another monitor and continued reading HN.

...30 minutes later the horse is still running and I'm like 'wtf? what does a horse have to do with the DDG logo?' close tab. read comments...

It turns out the app doesn't handle svg (it is actually in the to do list) and returned a 500, but the failure was never presented to the user.

Ouch. Sorry to have put you through that :/ I will fix it asap. And I agree with the parent too. Maybe some text below the horse that says fetching results... would make things clearer.

haha, no bother, we all know whats up here; I'm just busy bikeshedding.

Massive kudos for delivering; I like it a lot :D

Thank you! Fixed it :)

It would be cool if you could draw an icon instead of uploading/linking one, like how Google Translate lets you draw characters

Great idea. I'll definitely look into it. Thanks for the suggestion. Cheers

Nice idea! I did a similar but different project for 32x32 icons.

https://iconpush.github.io/

Please help yourself to my icon data - I spent a while collecting it and hope it can be useful to someone!

Thanks for the share! These icons are golden! I did something similar for logos that could be of interest to you - http://compute.vision/brands/colorpicker.html

Hi, how did you source the whole noun project icon collection?

I scraped it from their website and then asked for their permission by sharing this link with them. They appreciated that I linked all the icons to their website and gave their consent to make this public.

Ah, I see. I was hoping there was an archive to download somewhere. Thanks!

How do/would you measure the accuracy of the algorithm?

The MPEG-7 dataset[0] is what most researchers use to benchmark shape similarity algorithms. There are couple of other datasets that I used that I can't recollect now. These datasets are relatively simple with a single shape as opposed to logos, icons that comprise multiple elements in different configurations.

I would test on the MPEG-7 dataset to begin with and once the precision and recall values are good enough go ahead with testing on logos and icons. I must've manually tested the algorithm more than a 100,000 times probably because that was the only way to do with untagged datasets. Quite tedious indeed. This version gives out pretty decent results about 7-8 out of 10 times I'd say.

[0] - http://www.dabi.temple.edu/~shape/MPEG7/dataset.html

Traditional algorithms or ML? Image vectors?

Traditional algorithms and image vectors. I used a conconction of existing region and contour based techniques and threw in some original ideas as well.

Elaborate on some of your ideas or do a blog post?

Been meaning to do a blog post on this forever now :/ I will do it at some point for sure at http://linkdot.link

This is a great starting point in case you are interested in knowing more -- http://www.staff.science.uu.nl/~kreve101/asci/smi2001.pdf More recently, I've been exploring some ideas of Tversky.

One way is to scale the images down and threshold to black and white. At smaller scale they should be identical or much closer.

Reply on news.ycombinator.com