RGB-D Hough Forest

This is a MSR-SSLA licensed project for object detection in RGB-D images using Hough-Forest.

View project onGitHub

Welcome to RGB-D Hough Forest project page.

This code is developed during my master thesis in AIS, University of Bonn under the guidance of Prof. Dr. Sven Behnke

All directories contain CMakelists.txt, follow the instruction below to make the project

General steps to make your project using CMake.

  • In your project directory Make build directory using shell command mkdir build
  • go to the build directory cd build
  • ccmake .. will let you configure your make file.( The respective keyboard command options are shown at the bottom. )
  • make && make install
  • You can find your executable in the project folder.

For more information on how to run each code kindly read README.md file.

What you will find here?

  • Complete code for training the class-specific forest and 6-DoF object detection. (C++)
  • Already trained class-specific forests for 6 object categories (bowl, cap, cereal-box, coffee-mug, flashlight, soda-can).
  • Evaluation code. (C++)
  • Artificially generated training images from [RGB-D Obejct Dataset].(http://www.cs.washington.edu/rgbd-dataset/dataset/).
  • Code for generating training images. (C++)
  • Pose( only verticaly upward direction ) annotations for [RGB-D Scene Dataset].(http://www.cs.washington.edu/rgbd-dataset/dataset/)
  • Pose annotation code. (C++)

Authors and Contributors


Find answers to all your questions by contacting me on this E-mail id : badami@vision.rwth-aachen.de