gurpreetshanky / SfM

Structure from Motion pipeline implementation using RGB-D images

Geek Repo:Geek Repo

Github PK Tool:Github PK Tool

Structure from Motion Pipeline using RGB-D

A project by Seonwook Park and Yifan Wang for the 3D Photography course at ETH Zurich in Spring 2015.

Install

  1. Clone this repository
  2. Setup OpenCV 3, Ceres Solver and PCL
  3. mkdir build && cd build && cmake .. && make

Data Acquisition

  1. Descend into bernhard/
  2. Compile using mkdir build && cd build && cmake .. && make
  3. Plug a Microsoft Kinect (gen 1) in
  4. Run the binary from a directory in which data should be stored in. Use the flag -method 4 to store RGB and D images separately.
  5. While the application is running, press s to record single frames. Attempt to cover the target object or environment as much as possible.

Running the pipeline

  1. Get our dataset by running bash get_dataset.sh
  2. Descend into with cd build/
  3. Run the binary and provide the path to our dataset. For example, ./main ../dataset BeersNMore

About

Structure from Motion pipeline implementation using RGB-D images

License:MIT License


Languages

Language:C++ 57.7%Language:TeX 39.1%Language:CMake 3.0%Language:Makefile 0.2%Language:Shell 0.1%