Added a dedicated page for the CUDA RANSAC Homography code, under the Computer Vision section.
I’m in the process of restructuring the website a bit, moving some projects posted on the main blog page to dedicated pages.
I’ve updated the CUDA RANSAC Homography code. It now uses OpenCV’s SURF GPU implementation instead of FAST. The CUDA kernel based SVD code has been replaced by a port of GNU GSL’s SVD function. The CUDA SVD code is generic enough that you can easily include it into an existing project, but probably limited to small matrices. You can download it here CUDA_RANSAC_Homography_1.1.tar.bz2.
I’ve rolled out my own structure from motion package, called RunSFM, that aims to make it simple to get Bundler and CMVS up and running on Linux. And by simple I mean, a single make command to compile and a single call to RunSFM.sh to do the entire process, no arguments required. Okay, the catch is you do have to install some packages beforehand, all of which are on Synaptics in Ubuntu 11.04. The uncommon ones I’ve included for convenience.
RunSFM also includes speed optimisations I’ve made, like speeding up PMVS by up to 40% and parallel SIFT extraction and matching. No GPU is used, so most people should be able to use it. I decided not to use SiftGPU for now because I had a test case where it failed to find enough good matches compared to the SIFT binary by David Lowe.
You can check RunSFM out via Computer Vision -> Structure from Motion