Help Improve Robot Vision With Your Kinect

Discerning shapes and recognizing objects takes humans several years to learn and it is an essential skill to survive and interact with the world around us. Machines also need to learn this important skill and although some researchers have made progress in recognizing objects, shapes and faces, vision processing is still very complex and there is much more work to be done. Kinect@Home is attempting to contribute to the field by enabling everybody to create and share 3D scans of objects (like cats),  environments (bedrooms) and more. They hope their free online service will help gather lots of 3D data that robots could then use to learn to navigate and recognize everyday objects. Why is so much data needed? Usually, when it comes to machine learning problems, the more data is available the better and the faster a model can converge and produce decent results. [iframe  src="http://www.kinectathome.com/models/a510f08c-be61-4656-bcdb-10a3297c710c/?embed=1"] As you cans see above, they have a 3D data viewer that works in browsers (through flash) that can be used to share and embed scans. They also provide all the software needed to create those scans trough your browser using a Kinect (for windows only at the moment). http://www.youtube.com/watch?v=gvd95Z6AiuM This is the first time it has been so simple to make complex 3D scans of objects and environments, and we recommend everyone try this new service (especially if you own a 3D printer as well). Via Wired.