Robust Segmentation of cluttered scenes using RGB-Z images

With the availability of commercial depth sensors such as the Kinect, there is a body of work using the depth information provided by the Kinect to segment images of scenes [1][6]. In this project we try to segment cluttered scenes by augmenting color information in images with depth cues from the Kinect data. In this milestone, our major focus was to study the literature in the RGBZ domain, implement code to replicate the method in [4], and try different modifications in order to make the method perform better.