Posted in: Uncategorized

Early Thoughts on Riya Public Beta (Updated)

I have started to carve longer posts into short snippets and long body portions to make it easier to wade through some of these longer postings. I started playing with the Riya public beta last night and I am left feeling like I am missing something. I know that they are early on and are trying to build up the database of recognized faces and such, but my early thoughts are that the service asks a lot from me as a user and doesn’t necessarily give me a whole lot in return at the moment. I also worry about the privacy issues. That being said, I can see the potential for more value being created down the road.

I got a very nice email from someone at Riya clarifying a bunch of the questions that I raised — I am glad they found this post and were kind enough to respond. I think they have an interesting product and I poked it pretty hard because I think it has the potential to be really useful once it is integrated with more of my existing tools. It’s clear that they are working on a lot of the things that I want. The challenges of being an early adopter *sigh*.

I will keep an eye on Riya as it develops. It sounds like they are pointed in the right direction.

If you want to read more, just click on the continue reading more section. I wrote this late at night, so sorry if it rambles.

I have been eagerly awaiting the public release of Riya’s facial recognition product. To be honest, I was not sure what value it might create for me as a user or what unmet photo-related need it would solve, but I was really curious to see what the application would actually do. I have tried to organize my feedback around a few topic areas:

User Experience – Uploading Photos
I downloaded the Riya uploader to my PC. The download was really quick. If you have used any other photo uploading tool, the experience is quite similar — you point the uploader to a folder of pictures and tell it to do its thing. The Riya uploader performed slower than just about any other photo uploading app I have used — I was able to upload about 85-100 photos per hour. Based on the animations that I saw during the upload process (and my CPU utilization), it appears that the uploader does some pre-processing to focus on and identify things that appear to be faces in your photos before you upload them. I ended up uploading about 125 photos before I got bored and shut down the uploader.

White the experience was relatively painless, I am not really looking for YAPSPO (yet another place to store photos online). I use Flickr and Ofoto — I don’t want to have to park my photos in a 3rd place. It appears that an API is in the works to allow integration with other services. That will be a big help.

User Experience – Tagging Photos
I can see how the Riya team tried to make this process as painless as possible. The way the service basically works is that you can “train” the application to recognize particular faces by explicitly labeling them with names and email addresses. Once you have successfully identified someone, their name appears in a list on the left and you can drag and drop pictures on to names.

There were two parts of this process that I found remarkably tedious. First, Riya suggests that you take the time to tag every photo on the page before you click done. Some of my pages had 30-50 images on them; I didn’t want to sit there and tag all of them before clicking done. That was kind of annoying. Second, the service recommends that you re-tag each set of images 5x. Repeat the same set of actions on the same set of photos 5 times? Perhaps I misunderstood the directions, but that’s what it seemed to suggest. I tried this a few times, and every time I tagged some faces and clicked “Done” it generated a page with the same set of faces. The FAQ suggested that this is a known bug, but the little “Train, Rinse, Repeat” window was a bit ambiguous about what they really meant by repeat 5 times.

“Bang for the Buck”

After about an hour of tagging photos, I just stopped. I wasn’t really sure what the return was going to be for me as a user. All of the tools that I use to get photos off of my camera (I generally use Picasa, but don’t use it exclusively) allow me to dump photos in a folder. That more or less works for me. Many of the people who I photograph are known to me, so I don’t need facial recognition to identify them. Sure, I can see some value in being able to auto-tag photos with names, places, etc so that I can do post hoc searches (“show me all of the pictures with me and my family in 2005”), but I am not sure what I would do with the output of those searches.

That being said, many people find photos fascinating even if you don’t do anything except look at them. There are enough photo voyeurs on the web who just like looking at things that are cool or otherwise interesting (see Flickr) that I can see this kind of service generating a lot of goodwill tagging (people who tag just to help out, not to get any benefit). Until this is integrated with more of the photo services I use, I am not sure that the bang for the buck works out in my favor. For me, the basic tagging elements in Flick meet my needs. I have to imagine that the folks at Riya or some enterprising developer are hard at work on some nifty integration here.

Privacy Concerns
I am not a privacy zealot and I don’t think that the Riya folks have any nefarious aims. My larger concern, however, is how my friends feel about having me, as a user, upload their photos to a website where their face might be recognized by software. I have enough privacy-conscious friends who might feel this way that I stopped uploading photos to Riya after a bit. In the same way that it took awhile for people to figure out the appropriate ettiquette for social networking invitations (it might not be wise to invite your boss to view your crazy MySpace page, for example), it might take people awhile to figure out the proper social norms for Riya. I know a lot of people who would not want their photos posted on something like this. They don’t even like pictures from parties being circulated without their permission.

Also, because Riya has a fairly grand aim in what it wants to do with photos, it is unclear what applications will emerge to leverage the Riya API. What if those applications do annoying/embarassing things with photos or facial recognition? Not saying it will happen, but these are the kinds of things that came up in discussions I had with friends about this technology.

If Riya succeeds, it could be a cool and very useful service. For now, I think I will take a more wait-and-see attitude.

Comments? Email me.

Back to Top