VLFeat is a MATLAB-compatible implementation common computer vision algorithms such as SIFT, MSER, k-means, hierarchical k-means, and agglomerative information bottleneck.
VLFeat is easily portable and supports natively Mac OS X, Linux, and Windows.
VLFeat is open source software released under the GNU GPL V2 license.
You can obtain VLFeat source code or precompiled binaries from here. To start using the MATLAB toolbox, you can read some of the tutorials and have a look at the list of MATLAB functions.
Some algorithms (SIFT and MSER) have command line interfaces as well (which do not require MATLAB). To learn about the command line utilities please refer to the relative manual pages.
The core of VLFeat is a clean C library. To learn about the C API and the details of the implemented algorithms, please refer to the C library documentation.
If you use VLFeat in your research work, please consider citing us. If you use BibTex, you can use the following entry:
@misc{vedaldi08vlfeat, Author = {A. Vedaldi and B. Fulkerson}, Howpublished = {\url{http://www.vlfeat.org/}}, Title = {{VLFeat}: An Open and Portable Library of Computer Vision Algorithms}, Year = {2008}}