Deepfish - Enhanced Browsing on Mobile Devices

Deepfish is Microsoft's intelligent web browsing technology for mobile devices. It renders mobile web pages "in a view that is closer to the desktop experience". This kind of mobile Web navigation is very similar to the one shown by Steve Jobs in January at his demo of the iPhone. The Deepfish technology works on both cell phones and the pocket PC. But unlike the iPhone, the navigation is done via Zoom box - sort of like Google Maps, in that it allows the user to focus on a specific part of the page. [Ed: is this the same technology as ZenZui?]

Just by looking at the examples, we can tell that the scaling is not ad hoc - rather it is 'intelligent' and sensitive to phone size, colors and images on the page. Like iPhone, Deepfish also supports vertical and horizontal viewing mode, although this is probably dependent on the kind of device you are using. Deepfish works now on Windows Mobile 5 and can be downloaded from the live labs web site (also more info here). Overall it does not look as slick as iPhone, but it is certainly on a par with iPhone's technology.

Boku is a visual programming language aimed at teaching kids how to program. Boku looks like Microsoft Bob circa 2007, but it is actually quite cute - so kids will like it. The setup is that Boku is on a desert island and he is facing challenges. To help him solve these challenges, kids have to program it. The program renders in a realistic 3D landscape and has built in 'primitives' for sensing the surroundings and executing commands. Kids can program it by chaining together boxes.

For example, a chain might be: If you [see] [something red] [move towards it]. Seeing is a sensor function (like touch, smell, etc), something red is the filter and move is the command. The wide variety of sensors, filters and commands is what makes the world of Boku rich and interesting. 

To be fair, at first look it is no different in principle from the old MIT LOGO. But it certainly is flashy and video-game like, so it might be just right for kids.

Epitome - Image Feature Extraction Technology

Epitome is an interesting image compression technique, which creates a smaller image from the essential features of the original image. The 'epitome' of an image is its miniature, condensed version - containing the essence of the textural and shape properties of the image. It is based on probabilistic methods, which works by dissecting an image into a smaller set of patches. Each patch is of a different size and represents a piece of the original image. Feature extraction is then applied to create a small image that captures the "essence" of the original.

A really interesting example, that illustrates the application of this technology, showed a picture of 300 people. After creating an epitome of the entire set, the researcher pointed out a piece of the epitome, which represented all smiles found in the original image. By clicking on the smile, he was able to retrieve all people who were smiling in the original picture. You can learn more about this technology, as well as see some examples, on the project web site.