Show HN: How a machine learning model understands logo similarity
Hi HN, I'm a co-founder of Aesthetic, a new kind of research tool for designers.
Our upcoming app lets designers search the world of design using images or text. To understand and improve the quality of our visual similarity models, we want a more intuitive way to interpret some of the distances between our logo corpus.
My co-founder Andrew Look created this awesome tool using Tensorflow, Inception, t-sne and a subset of our logos corpus.
We just published a cool video of it, published it our website. See it here: www.aesthetic.com
And, here's another data viz that we didn't post on our landing page: https://gph.is/g/aNWDeoq
Posts without URLs get penalized, so you should make a new post that points to your site, then post your text as a first comment in the thread. If you do, email hn@ycombinator.com so we can make sure that it doesn't get flagged. You might also include the backstory of how you came to work on this and explaining what's different about it. That tends to seed discussion in a good direction.
Edit: I took a quick look at www.aesthetic.com and it appeared to be a signup form rather than something users can directly try out now. If that's the case, it would be better to wait before reposting your Show HN: see https://news.ycombinator.com/showhn.html.