The Berkeley Segmentation Engine (BSE)

contact: charless fowlkes

developed in collaboration with david martin, and jitendra malik

This code release contains libraries for computing pairwise affinities with the goal of segmenting natural images. There is also code which feeds the computed affinities to an eigensolver (TRLan) and clusters the resulting eigenvectors using k-means to yield an image segmentation.

The algorithms used are described in:

  1. C. Fowlkes, D. Martin, J. Malik. "Learning Affinity Functions for Image Segmentation: Combining Patch-based and Gradient-based Approaches", CVPR, Madison, WI, (June 2003). [pdf]

  2. D. Martin, C. Fowlkes, J. Malik. "Learning to Detect Natural Image Boundaries Using Local Brightness, Color and Texture Cues", TPAMI 26 (5) p.530-549 [pdf]

  3. C. Fowlkes, J. Malik. "How Much Does Globalization Help Segmentation?", Technical Report CSD-04-1340, Division of Computer Science, University of California, Berkeley, (July 2004). [pdf]

This codebase will likely undergo some major revisions as more segmentation functionality is added. Please email me if you would like to recieve an announcement of the next release.