Autonomous Space Robotics Lab

Speeded Up SURF

UTIAS ASRL

Build instructions for OSX

1) GET MACPORTS

We use macports to get all of our dependencies.

http://www.macports.org/

2) GET CUDA

Download and install CUDA 2.3

http://developer.nvidia.com/object/cuda_2_3_downloads.html

You will need

  1. the development driver
  2. the CUDA toolkit and
  3. the CUDA SDK code samples

3) PICK THE RIGHT COMPILER

In Snow Leopard, we had to pick gcc 4.0 as CUDA 2.3 only exists as a 32bit library. Open a console and type

  sudo port install gcc_select
  gcc_select gcc40

4) GET OpenCV, CMAKE, and Boost

Here is where macports comes in handy. Open up a console, type

  sudo port install cmake opencv boost

wait.

On Snow Leopard we had some problems between the architecture of opencv installed by macports, and that needed by gpusurf. To resolve this, we compiled opencv from source.

5) COMPILE CUDPP

The segmented scan from the CUDPP project is used to build the integral images.

http://code.google.com/p/cudpp/

This project is not in macports but we have included it in the 3rdParty directory. Open a console and navigate to 3rdParty/cudpp/cudpp and type

  make

wait.

6) BUILD GPU SURF

open a console in the build directory. type:

  cmake ..

This gets cmake to construct the makefile. Now you can build the library. Type:

  make

This will produce a few things:

  1. The bin/gpusurf_engine command line utility.
  2. The bin/gpusurf_tests command line program.
  3. The lib/gpusurf.dylib dynamically linked library file.

Generated on Fri Apr 30 20:06:20 2010 for gpusurf by doxygen 1.6.2