Home

The VLFeat open source library implements popular computer vision algorithms including SIFT, MSER, k-means, hierarchical k-means, agglomerative information bottleneck, and quick shift. It is written in C for efficiency and compatibility, with interfaces in MATLAB for ease of use, and detailed documentation throughout. It supports Windows, Mac OS X, and Linux. The latest version of VLFeat is 0.9.9.

Download

Documentation

Tutorials

Example applications

Citing

@misc{vedaldi08vlfeat,
 Author = {A. Vedaldi and B. Fulkerson},
 Title = {{VLFeat}: An Open and Portable Library 
          of Computer Vision Algorithms},
 Year  = {2008},
 Howpublished = {\url{http://www.vlfeat.org/}}

Acknowledgments

UCLA Vision Lab, Oxford VGG.

News

14/06/2010 - VLFeat 0.9.9 released
VLFeat 0.9.9 adds a new sample application (SIFT matching) and minor refinements. [Details].
14/06/2010 - Open Source Vision Software Tutorial
VLFeat presented at the CVPR 2010 Open Source Vision Software Tutorial. Slides of the presentation are available from the tutorial web page.
10/05/2010 - VLFeat 0.9.8 released
VLFeat 0.9.8 adds new tutorials, (hierarchical) k-means support for floating point data, homogeneous kernel maps, a basic implementation of PEGASOS for SVM learning, and many other improvements. [Details].
16/01/2010 - VLFeat 0.9.7 released
VLFeat 0.9.7 updates the binary distribution to be backward compatible with Mac OS X 10.5 (Leopard). [Details].
10/01/2010 - VLFeat 0.9.6 released
VLFeat 0.9.6 contains minor improvements to the binary distribution. Specifically, it makes VLFeat GNU/Linux distribution compatible with the older GLIBC version 2.3. [Details].
30/11/2009 - VLFeat 0.9.5 released
VLFeat 0.9.5 adds a fast kd-tree implementation and SSE-acelerated vector/histogram comparison. The dense SIFT (dsift) implementation has also been improved. Binaries and compilation support for Mac OS 10.6 (Snow Leopard) and MATLAB R2009b (32 and 64 bit) have been added [Details].
MATLAB 7.0 and earlier require recompling the MEX files by the provided vl_compile command.

Acknowledgments

Part of this work was supported by the UCLA Vision Lab and the Oxford VGG Lab. The authors would like to thank the many colleagues that have contributed to VLFeat by testing and providing helpful suggestions and comments.