A just-launched visual search site, Ave23, has an appealing concept: drag a photo of an item or outfit you like into a search bar, hit enter, and watch the page populate with similar garments and accessories from a variety of brands at different price points. In theory, the platform would allow users to customize their wardrobe with pieces that appeal specifically to their own style, based on high-fashion and celebrity reference points.
Here's the hitch: it doesn't really work. I consider myself to be pretty computer savvy, and I, for the life of me, couldn't figure out how to upload photos from other sites to the Ave23 database. Other users had populated the site with celebrity style images, so maybe I'm just dimwitted today — but either way, this is a major usability issue. After all, Google never needed an FAQ or user manual to teach us how to use the search box.
All things being fair, Ave23 is in beta, and visual information is a lot harder to index than text, so I'm willing to give the site a little leeway. I'd understand if the engine wanted to roll out broader search functionality over time, and only allow users to play with the items pre-loaded onto the site, but even then, Ave23 doesn't do a good job of finding similar pieces for the items you can search. I pulled up a street style look that had been uploaded to the site's community pages, and did a sample search. A casual, unstructured brown blazer pulled up a slew of tailored, dressy, black blazers: I guess it's impressive that the technology can recognize that the garment hanging on a woman's shoulders is a dark-colored blazer, but if it can't make more precise distinctions, it's not a useful tool.
The search engine was just launched today, so some of these issues could be due to server overload or other short-term tech issues. I really do hope Ave23 improves, but the shortcomings of the site have already been solved by another entity: us.