Electrical Engineering
      and Computer Sciences

Electrical Engineering and Computer Sciences

COLLEGE OF ENGINEERING

UC Berkeley

Object Detection in RGB-D Indoor Scenes

Edmund Shanming Ye

EECS Department
University of California, Berkeley
Technical Report No. UCB/EECS-2013-3
January 14, 2013

http://www.eecs.berkeley.edu/Pubs/TechRpts/2013/EECS-2013-3.pdf

Object detection is a well-studied problem in computer vision. One of the basic tasks is to draw tight bounding boxes around instances of various target classes in a set of images. Computer vision literature has primarily focused on intensity, with less emphasis on depth data. In this report we address the challenge of detecting 10 common household items (bed, chair, etc) in RGB-D images obtained using the Kinect. We operate on the recently released NYU-Depth V2 dataset. Our algorithm augments the deformable parts model by adding a set of vector quantized depth features that are, to the best of our knowledge, novel on this dataset.

Advisor: Jitendra Malik


BibTeX citation:

@mastersthesis{Ye:EECS-2013-3,
    Author = {Ye, Edmund Shanming},
    Editor = {Malik, Jitendra},
    Title = {Object Detection in RGB-D Indoor Scenes},
    School = {EECS Department, University of California, Berkeley},
    Year = {2013},
    Month = {Jan},
    URL = {http://www.eecs.berkeley.edu/Pubs/TechRpts/2013/EECS-2013-3.html},
    Number = {UCB/EECS-2013-3},
    Abstract = {Object detection is a well-studied problem in computer vision. One of the basic tasks is to draw tight bounding boxes around instances of various target classes in a set of images. Computer vision literature has primarily focused on intensity, with less emphasis on depth data. In this report we address the challenge of detecting 10 common household items (bed, chair, etc) in RGB-D images obtained using the Kinect. We operate on the recently released NYU-Depth V2 dataset. Our algorithm augments the deformable parts model by adding a set of vector quantized depth features that are, to the best of our knowledge, novel on this dataset.}
}

EndNote citation:

%0 Thesis
%A Ye, Edmund Shanming
%E Malik, Jitendra
%T Object Detection in RGB-D Indoor Scenes
%I EECS Department, University of California, Berkeley
%D 2013
%8 January 14
%@ UCB/EECS-2013-3
%U http://www.eecs.berkeley.edu/Pubs/TechRpts/2013/EECS-2013-3.html
%F Ye:EECS-2013-3