/* ---- Google Analytics Code Below */

Wednesday, July 24, 2019

Component Identification, Learning

Quite interesting project.  A way to further connect data to parts and then wholes?    A kind of basic semantic representation?     Can think of a number of ways this could produce much value.

Intel Does the Hard Work so Robots Can Operate Your Microwave 
Tech Crunch
By Darrell Etherington

Intel artificial intelligence (AI) researchers, in partnership with the University of California, San Diego and Stanford University, have compiled a large dataset of three-dimensional objects featuring highly detailed, hierarchically structured and fully annotated information. The PartNet dataset organizes objects into segmented components, in a manner applicable to building AI learning models, for identifying and manipulating actual objects. PartNet lists more than 570,000 components across more than 26,000 objects, with parts common to objects across categories labeled as corresponding to one another. This enables AIs taught to recognize a part on one object variant, to identify it on another.   ... "

No comments: