Showing posts with label facial recognition. Show all posts
Showing posts with label facial recognition. Show all posts

10.21.2009

A Huge Day for the Big Web Players


It's been a huge day for the major web companies, news pouring out every hour, so I thought I'd recap and offer some thoughts:

Music
  • Facebook users will soon be able to gift each other songs, web-streaming copies for 10 cents or downloadable DRM free files for 90 cents, powered by lala.com (paid thru Facebook credits)
  • Google is expected to unveil a music tie-in next week that lets users stream songs and potentially purchase them right from the search engine, powered by i-Like and lala.com
  • MySpace will now be housing a full library of music videos from the major music labels, and also providing Artists with deeper analytics of how fans engage with their content
Clearly this all points to one thing - multimedia content is rapidly becoming available everywhere. The entertainment industry is learning the only way they can fight piracy to offer easy access at fair costs. Also, I think the ability to gift Facebook friends songs could be a huge money maker for all parties if it's easy for the user and the music is easily played back as a library. And I want to see if users buy their own songs to house on their own profile for easy listening. Also, its amazing a full mp3 is cheaper than the virtual gifts people can send now. Finally, it's impressive that Lala was able to get themselves into both the Facebook and Google music deals, taking them from relatively no-name to top-tier player in a matter of hours.
Myspace's announcement is them trying to hold onto their claim as the best place for Musicians to set up their home-base website as they watch more and more of the grassroots marketing move to Twitter & Facebook. They're doing a pretty good job as an Entertainment portal, but this helps them with actual artist relations.

Social Search
  • Microsoft cut deals with Twitter and Facebook to incorporate their content streams in real-time into their Search index. Microsoft unveiled an early beta of their Twitter search engine at http://www.bing.com/twitter , which not only shows recent updates but popular links being shared around a search term
  • Google announced a similar deal with Twitter (no Facebook for now), but hasn't yet incorporated the content fully, with more details expected in a few weeks
  • Google announced a separate product called Social Search which will enable users to search their own social graph for content. Google will create a personal index of content for a user based on the social profiles attached to their Google profile, helping users surface content such as friend's Flickr photos and profile info
The social search space, also known as the real-time search space, is heating up big time. The major search engines Google and Bing are realizing that more and more users will start their web searches with their social graph and want the latest information that their social graph is highlighting as valuable. Microsoft's early search site shows how it will try to take on the memetrackers and the real-time search engines at once, a space that's currently being owned by new web start-ups Tweetmeme and OneRiot.
It's clear Google will jump into this space as well, but their personalized social search product is separately very interesting. Helping users to unearth their own personal information from their own individual social graph is potentially even more important than a broad real-time search engine. The key here though is what content Google will be able to index - given that Facebook especially is behind a wall. The great thing about FriendFeed was that you could add RSS feeds to supplement the content that was already being indexed, a way to overcome the Facebook walls. Will Google be able to do this?

Photos
  • Flickr announced people tagging on their photos similar to Facebook's features, but with more privacy controls. Users can choose who exactly can tag them in photos.
This is a feature that's made sense for a long time, but I don't think by itself it will matter much. Not enough people have Flickr profiles for it to matter, and because of the upload caps not enough people will ever choose to upload to Flickr. If this was done through Facebook connect it would be totally different (similar to PolarRose's offering). Also I would really like to see some integration with Apple's iPhoto, since they are working on syncing their products already and I've invested the time in tagging my friends on my desktop computer. This would be a good rivalry to the sync'd tagging of Picasa and Picasaweb. Finally, since serious Flickr users have been stockpiling photos for years, Flickr really needs to integrate Facial recognition on the site - both to keep up to par with competitors, and to make tagging less daunting of a task.

6.01.2008

Image Recognition Ads Come to Facebook

I know this seems kind of creepy, but I can't help be incredibly impressed by this: Like.com, which formerly used to be Riya.com, the facial recognition photo sharing service that I was obsessed with, but then changed over to an object recognition shopping site (show us an item that a celebrity is wearing and we'll show you 50 things just like it ranging in all types of prices), is now running ads on Facebook that match the picture in the ad with your profile picture. TechCrunch has an example where the ad features a guy wearing aviator glasses because Riya's technology recognized aviator glasses. This is very cool, albeit a bit creepy. Note that TechCrunch is not confirming this yet, just suggesting based on the technology Riya is known for that this type of image recognition match is happening in the background - so maybe I should hold off a sec before I get too excited.

4.18.2008

Riya - Please sell your facial recognition technology to someone who will use it

Riya, the photo-sharing website with facial recognition capability that I've obsessed over and written often about, is looking to sell their facial recognition technology. They transitioned over all of their efforts to a visual shopping search engine which is apparently very lucrative for them, although not nearly as exciting or fun. Apparently they never got over the hump in making the facial recognition software as accurate as they wanted, and any talks with buyers of Riya in that incarnation died long ago. Now they will officially rename their company Like.com, and sell off the Riya technology. I am hoping that Google or Yahoo purchases them and implements the technology immediately into Picasa or Flickr respectively - it would be an invaluable asset in tagging the thousands of photos I now have in each service (my wish list also includes those two companies allowing inter-operability & sync'ing). But now I'm thinking more likely someone like Facebook buys the technology and uses it to help its users tags the faces of their friends, a practice that Facebook users are already doing and are used to, and probably wouldn't mind the help.

9.19.2007

Viewdle is Facial Recognition in Videos

New video application Viewdle is actually able to pull off significant facial recognition within video.  It takes one detailed image of a person's face and then picks out when the person appears in video.  This leads to deep tagging and rich video search.  I'm not sure whether this will be available for personal use (could they offer a browser extension where this is utilized in video we've already uploaded on YouTube, Google Video, and Facebook), but it's alive and impressive on the Reuters, allowing you to search through their video archives.

Other recent articles about facial recognition in images: Riya 1,2 (my favorite), Google, AOL

12.31.2006

AOL Testing Facial Recognition Image Organizer for Your Photos









I haven't read anything else about this, but AOL has a prototype of a facial recognition image organizer available for download at their beta site, AOL Greenhouse. It has a fairly poor interface, so once you start to tag your photos going back and viewing them is, say, not even close to as easy as with Picasa. Still, it does identify faces in pictures pretty well, groups similar faces, and is fairly intelligent in tagging people once you've begun training it. In early testing, it seems to be more accurate than Riya, but isn't even as usable as Riya's web interface.



The problem with this, of course, and the problem with Riya, Flickr, Picasa, etc, is that they all have their good and their bad, but none of them are as strong as I want them to be. Picasa is necessary for managing a full collection of photos on your desktop, but Picasa Web Albums is limited in storage and usability, so I need to have Flickr on the web. Flickr is a step behind in web functions, though, and could definitely benefit from the addition of some AI such as facial recognition.



So now I have photos in Picasa, in AOL Image Organizer, on Flickr, on Riya, on Picasa Web Albums, and a few scattered on things like Zooomer. What I really want is the ability to share information from site to site - share geotagging, facial tagging, descriptions, etc, and important/sync the data across the storage / community sites. It simply takes too much work to manage a large collection of pictures at this point, and no one property is strong enough to handle all of my wants.



So is anyone else writing about AOL Image Organizer? Is anything out there going to tackle these problems?