VLFeat -- Vision Lab Features Library

Version:

0.9.14

Author:

Andrea Vedaldi (vedaldi@robots.ox.ac.uk)

Brian Fulkerson (bfulkers@cs.ucla.edu)

Copyright © 2007-12 Andrea Vedaldi and Brian Fulkerson

VLFeat C library contains implementations of common computer vision algorithms, with a special focus on visual features for matching image regions. Applications include structure from motion and object and category detection and recognition.

We strive to make the library free of clutter, portable (VLFeat is largely C-89 compatible), and self- documented. Different parts of the library are weakly interdependent, simplifying understanding and extraction of code.

Contents

• Algorithms
  • Scale Invariant Feature Transform (SIFT)
  • Dense Scale Invariant Feature Transform (DSIFT)
  • Maximally Stable Extremal Regions
  • K-means clustering
  • Integer K-means (IKM)
  • Hierarchical Integer K-means (HIKM)
  • Agglomerative Information Bottleneck (AIB)
  • KD-trees and forests
  • Homogeneous kernel map
  • PEGASOS SVM solver
  • Simple Linear Iterative Clustering (SLIC)
  • Quick shift image segmentation
• Support functionalities
  • Platform abstraction
  • Preprocssor, library state, etc.
  • Random number generator
  • Math operations
  • Generic heap object (priority queue)
  • String operations
  • Image operations
  • PGM reading and writing
  • Rodrigues formula
  • MATLAB MEX helper functions
  • Drop-in getopt_long replacement
• VLFeat design concepts
  • Objects
  • Memory and resource management
  • Multi-threading
  • Portability
• Glossary