Subscribe
Sign in
LikeThat Style — Shazam for fashion
21
Replies
Orli Yakuel אורלי
Can't wait to try this!
Share
we want to use the API but we want the search hot the shops that listed have our
link , can be that possible ?
@ilyashassani - yes, our API can be used for custom indices. You can email api@justvisual.com, and I'll also DM you with my own email.
@rish_says - thanks for trying it out, interested to get your feedback. Let your friend know we have an API if they want to use our search tech - just go to our website (justvisual.com).
@kevinakiralee It worked smoothly for us. Interestingly, we tried a "Short Kurta" worn by a friend and indian fusion garment which I thought would be difficult to detect, it did a pretty good job. Then we tried a grey regular Adidas sport men's T-shirt on a fiend- while it gave decent suggestions - it gave both men and women which was acceptable. What was weird was the first option was a red T shirt even though next 5 were grey. That was a bit off - first T shirt shown being red. But it works well overall, impressive.
@rish_says - thanks for the detailed feedback, very helpful. The app is meant to focus on women's clothes, but since we crawled hundreds of stores to get to our 10M+ item catalog, some men's items have snuck in. We're also working to further improve the precision - our own tests show that we usually have excellent (often exact) hits in the top 10-15 results, but too often the best results are not in the top 3. It's a work in progress!
@ilyashassani We don't have a specific API for men's style, but there's a pretty good chance that the training will work well when applied to a men's catalog.
Hi Product Hunt!
I'm the head of product for JustVisual, a company that's built a visual search engine. Our latest app is called LikeThat Style. It's an app for searching and exploring women's fashion, based entirely on images. You can use pictures from your camera/roll, or even pull in screenshots from Instagram or other social media apps. We then find matching and similar items that are shoppable and in-stock.
While other apps have attempted this, we've made the biggest investment in deep learning, so our visual relevance is much better. In particular it works with 'natural' pictures like the ones on your camera roll or Instagram, not just clean catalog images. While the technology isn't perfect, we believe it's now developed enough to be genuinely useful to the average user.
We've also spent a ton of time with UX design and research to make the app easy to use. While this is an early version, it still reflects multiple iterations on tuning the first-time user experience to communicate the concept of visual search. We're currently seeing 90% of users successfully do a search. So PH'ers might also find it interesting as an example of onboarding for a relatively complex product concept.
Thanks for checking it out, and we're very interested to hear your thoughts!
@kevinakiralee Is the API currently available? I could see great integrations into our product at divvi.
@swobos - yes, API currently available, and free to use! Info at justvisual.com, and I'll DM you with more details.
love the concept Kevin . I think since you have opened your API the possibilities of this tech are immense and I'm sure well be seeing a lot of interesting ideas come alive through this
@kevinakiralee Just tried your app and it is really cool. It is automatically scanning and searching for similar items, when Wheretoget app is working collaborative thanks to the tips and recommendations of other users to find the exact same item. You should really work together!
@emmanueldarmon - thanks for the tip. We're focused on the core technology, and I agree that there are some interesting possibilities with integrating it into a bunch of different products/contexts. We actually are offering an API to do just that - it provides access to the same core search tech that powers the app.