Jack000

41 ms·

391.
by Jack000 9y ago
thanks, I'll look into it
392.
by Jack000 9y ago
does the page load? more details would help me identify the issue
393.
by Jack000 9y ago
at least I didn't call it AI :]
394.
by Jack000 9y ago
server's at 90x but uploads do complete eventually.. just really slow right now from hn traffic!
395.
by Jack000 9y ago
maybe "basic computer vision"? dilate/erode/open/close filters come from computer vision afaik.
396.
by Jack000 9y ago
what seems to be the problem? server getting hammered by hn traffic might have something to do with it
397.
by Jack000 9y ago
oops, fixed!
398.
Show HN: Logo Crunch – Multi-resolution logo maker(brandmark.io)
328 points by Jack000 9y ago|65 comments
399.
by Jack000 9y ago
it would be cool to have this with a more neutral data set like wikipedia or google news. I think the biggest problem with word2vec is that it only captures the context of the word and not the meaning. eg. the top results for "cold&quo
400.
by Jack000 9y ago
That sounds more like a bin-packing problem than a jig-saw problem. If the irregular pieces fit in a specific way the problem is simplified (just check interfaces), if they don't the problem is much harder as you'd have to calcula
401.
by Jack000 9y ago
have you tried the sweep? I cancelled my pre-order after seeing some bad reviews
402.
by Jack000 9y ago
what's the issue? I'll have a look
403.
Show HN: Logo Crunch – Make your high-res logo legible at lower resolutions(brandmark.io)
9 points by Jack000 9y ago|2 comments
404.
by Jack000 9y ago
I hated it at first but started liking it after a few weeks. 9 months later I kind of hate it again. You get used to the short travel, and the clickiness is very pleasant. The problem I have now is that sometimes you can feel a click and it
405.
by Jack000 9y ago
if the tool told you something you already knew - that notable logos are good, what would be the point? maybe there's more value in discussing why the algorithm thinks bland logos score higher than the notable ones.
406.
by Jack000 9y ago
see thenounproject.com
407.
by Jack000 9y ago
someone's been abusing the server.. time to put up a captcha I guess :/
408.
by Jack000 9y ago
deleted after a day
409.
by Jack000 9y ago
yeah it doesn't really work for word-based logos. It's based on thenounproject, which has mostly square-ish icons.
410.
by Jack000 9y ago
one or two exact-matches doesn't affect the score much. It considers the aggregate similarity of the 150 most similar icons, out of 1M icons. In Apple's case it's mostly because there are a lot of apple icons that look very s
411.
by Jack000 9y ago
maybe that image really is a better logo than nike :] - on a uniqueness front there are a lot of swooshes, peppers and moustaches that look sort of similar to nike, whereas that image is guaranteed to not have been used for any icon or logo
412.
by Jack000 9y ago
it's definitely not perfect. I think the embedding-vector approach of image search isn't scale invariant.
413.
by Jack000 9y ago
hey guys this is the same neural net I use for the logo generator on brandmark.io - it basically does a visual similarity search on thenounproject to see how much your logo stands out from stock icons. it's intended to rate the technic
414.
Show HN: A neural net that critiques your logo(brandmark.io)
152 points by Jack000 9y ago|61 comments
415.
by Jack000 9y ago
I've ridden a boosted in NY and own two Chinese boards. I think the biggest problem with Boosted is that the vision they sold in their TED talk isn't very realistic. The main issues I have: - they need bike lanes. You could ride i
416.
by Jack000 9y ago
there was a tax issue in our shopping cart software that only occurred in florida. I googled for an address in florida and put in a few test purchases. Long story short, I shipped a thousand dollars worth of networking equipment to Disney W
417.
by Jack000 9y ago
so all you have to do is register? Is everyone who apply approved? they make it sound like an exclusive thing by leading with that line.
418.
by Jack000 9y ago
"Are you a member of the NVIDIA Developer Program in the US or Canada? If so, you’re eligible for an exclusive developer discount on the NVIDIA® Jetson™ TX1 Developer Kit SE" is it easy to join the developer program? Seems like no
419.
Show HN: Critiquing logo designs via deep learning(brandmark.io)
3 points by Jack000 9y ago|1 comments
420.
by Jack000 9y ago
wait, wouldn't scaling income by a constant multiple be the same as not changing anything? Your income and expenses are both doubled. this would just kill the bank accounts of anyone holding cash.

More ›