3D Model Based Object Recognition

Aleksandrs Ecins

The aim of the 3D model based recognition is to detect and estimate the 6DOF pose of a specific object in the scene. The input data is RGB-D data acquired with a Kinect sensor. For this project we were trying to recognize wooden planks. The size of the plank was 0.14*0.4*0.01m. Due to the noise in the depth sensor it was decided to model the plank as a 2D rectangle as opposed to a rectangular cuboid.