image detection

[Tatan]

Google’s hunger of acquisition is not news for anyone these days. There are fresh ones that will (or will not) turn into something (hopefully) profitable for us as consumers and there are some older ones that already are giving some fruits.
Let’s have a look at the first results from an acquisition of Neven Vision, a company specialized in image recognition. Some time passed and Google Image Search evolve up to the degree that it can now tell if the image was a picture of Paris Hilton or if it is just a picture of a lovely French city called Paris. It is just an example how “smart “ Image Search is getting .

What we can see at the present moment is the only faces and news-related images categories. Other categories are on the way.


The post can be found here :
http://googlesystem.blogspot.com/200...-to-faces.html

What issues in image detection the company is facing ?
How far can the Google go with that? What evolution can be predicted here?
How would the Google profit from this?

[bronz]

What issues in image detection the company is facing ?
How far can the Google go with that? What evolution can be predicted here?
How would the Google profit from this?

Issues: Sounds to me like an area ripe with technical issues. Are images manually tagged and indexed or crawled? What images are crawled and what are overlooked, given Google's penchant for ignoring porn sites? How about low-res images? Also, if image is going to be a niche Google wants to develop, how about other media? Music, video, sound, etc.? I'd love a Search Engine to find me user names, or a melody, or a tune on the radio. Image search also seems to me to be extremely problematic unless Google develops some rudimentary AI to filtre out extraneous results.

Evolution: VIDEO. An image may be worth a 1000 images, but a video is 24 images a second. That's a lot of words. How about making elements of these videos clickable, much the same way text can be monetized with contextual advertising? Imagine you see a video of a dude wearing cool sunglasses - you click on that image and you're taken to Ray Ban's site. Seems to me like a prime target area for advertisers.

Profit: ADVERTISING (geez, all my paragraphs are seguing into the next ones). Also advertisers to bid on specifc queries particuliar to a media type. Sunglasses are a good example, but how about hotels in e-cards? Paris Hilton's face leading one to her infamous video? Lots of spam potential, unfortuneately, as it's easier to distribute pictures and set up doorway and useless sites filled with them than it is quality text.

In fact, this might really influence SEO if Google can decipher image elements - what about them being able to decipher Flash, Java and text within image? Something to think about