Vision

The RVC Vision can be divided in two use case based sets of components:

• Dynamic Vision

• 2.5D Vision

The former is actively tracking a set of object based of a Deep Neural Network (DNN) in 2D world-space and then feeding the results to a second stage sub-component deriving the 3D space 6DoF pose using a depth based stream from the camera algorithm based on Pointcloud object matching

The latter is using Computer vision 2.5D space recognition based solely on RGB stream of a camera to derive a non real time pose of an object standing still a fixed distance surface.