addressalign-toparrow-leftarrow-rightbackbellblockcalendarcameraccwcheckchevron-downchevron-leftchevron-rightchevron-small-downchevron-small-leftchevron-small-rightchevron-small-upchevron-upcircle-with-checkcircle-with-crosscircle-with-pluscrossdots-three-verticaleditemptyheartexporteye-with-lineeyefacebookfolderfullheartglobegmailgooglegroupshelp-with-circleimageimagesinstagramFill 1linklocation-pinm-swarmSearchmailmessagesminusmoremuplabelShape 3 + Rectangle 1ShapeoutlookpersonJoin Group on CardStartprice-ribbonShapeShapeShapeShapeImported LayersImported LayersImported Layersshieldstartickettrashtriangle-downtriangle-uptwitteruserwarningyahoo

SF Bay Area 3D Vision & Kinect Hacking Message Board SF Bay Area 3D Vision & Kinect Hacking Discussion Forum › Kinect Fusion is a game changer for hard AI problems

Kinect Fusion is a game changer for hard AI problems

A former member
Post #: 1
Kinect Fusion. How it works and how it is a game changer:­

Hi all!

This video was designed for a less technical audience to help get a general overview of how Kinect Fusion and Kinect work.
I always find it hard to understand computer vision algorithms and I want to make it easier and more exciting for new people that might be interested in the field, newbies or maybe for people that have never heard of kinect fusion.
I also think Kinect+Kinect Fusion+CUDA are fundamental game changers for the development of AI or AGI. That's another reason I made this video. But, I hope some of you will find it useful and it will get more people involved in further development of PCL's version of Kinect Fusion. I certainly plan to contribute if I can.

I think that the technology in Kinect Fusion, Kinect and CUDA is so pivotal and revolutionary that it will allow us to solve some of the hardest AI problems between 5 and 10 years from now.
Some of the problems I think it will help solve are:
1) Gathering extensive and detailed knowledge about the world for use in any AI algorithm you could think of. This has not been possible until now.
2) Being able to deeply understand and interpret 2D images including stereo images with precision that is close to or as good as human capability. This can be accomplished by collecting massive amounts of training data which can then be used to develop better algorithms and even knowledge-based algorithms that know the behavior and/or 3D model/description of objects. Imagine for example that you wanted to create an algorithm that could easily determine whether a darker section of an image is a shadow or not. Knowledge about the object that the section likely belongs to or about previous shadow training data can make it possible with very high accuracy.
3) Natural language understanding and natural language learning; Natural language data mining
4) Search engines that mine data and answer questions directly.

Those are just a few examples I can think of at the moment. The list of applications could go on and on. I think this technology will radically accelerate progress in artificial intelligence.
Powered by mvnForum

People in this
Meetup are also in:

Sign up

Meetup members, Log in

By clicking "Sign up" or "Sign up using Facebook", you confirm that you accept our Terms of Service & Privacy Policy