After realizing last night that running the Riya uploader on a laptop is not the best idea, I installed it on my main computer at home, with all of my digital photos, and told it to start uploading. More than 6000 photos, and 24 hours later, it’s about 70% done.
But of the hundreds of faces it’s already isolated, I’ve been playing around with the identification training, telling Riya who the faces are, and I have to say, this is some pretty freaking cool stuff. 🙂
You can do a search within your photos, your Friends’ photos, and all public photos for a particular person’s face. You get an RSS feed to subscribe to for those search results – do a search for yourself, subscribe to the RSS feed, and now you’ll know whenever someone posts a photo of you to Riya. Pretty mindblowing. 🙂
You can also search for photos containing text, but I’m not sure if you can search for particular text. I’ll have to play with it some more.
I’ve done about a dozen rounds of the manual training – you’re presented a grid of your photos, and then can tab through an AJAX-y autocomplete interface to very quickly and easily tell it who each face is. When you’re done with a round of this (Riya recommends doing this at least 5x, but it’s addictive, and fun! :-)), it searches through your photos with the new names you just gave it, and shows you some proposed matches. You tell Riya which of the guesses are NOT the right person. It’s pretty dead on most of the time.
At any rate, I’ve got a lot more playing with this to do, but I can’t wait to have a way to search for all of my photos based on who is in them. If you’re like me, and don’t rename your photos after they come out of the camera, this is like a magic app that will parse through and tag your photos, kind of like what Musicbrainz does for MP3 files and filling in ID3 tags.