STAIR Vision Library (v1.1) Design and API Documentation
Overview
The
STAIR Vision Library (SVL) is a research and development
library focused on machine learning applications in computer
vision. This is reflected in the project directory structure which is
aimed at allowing multiple student groups to work on different
projects simultaneously while sharing the core library code. It is
also reflected in the
coding guidelines.
The development software repository is held at Stanford University and
is available only to Stanford students. Official software releases,
available to everyone, occur regularly on this website.
Directory Structure
The following represents the Stanford University STAIR Vision project
repository. The STAIR Vision Library (SVL), made available
publicly, represents only part of this directory tree.
Directory | Contents |
bin¹² | Compiled SVL libraries, applications, and projects |
data¹² | (empty) Recommended for storing or linking to datasets |
doc | SVL documentation (static copy of http://ai.stanford.edu/~sgould/svl/) |
experiments¹² | (empty) Recommended for running experiments |
external | External libraries required by the STAIR Vision Library. Small libraries are included with the SVL source code. Larger libraries (OpenCV and wxWidgets) should be downloaded separately and placed (or linked) here. Under Windows, OpenCV and wxWidgets should be installed in their default system locations. |
include | Include directory for applications that use the SVL |
models¹ | Recommended for storing model files, e.g., for object detection |
projects¹ | Recommended for STAIR Vision group projects |
sandbox¹ | Individual user directories for prototypes and throw-away code |
svl/apps | STAIR Vision Library applications source code |
svl/lib | STAIR Vision Library libraries source code |
svl/scripts | STAIR Vision Library perl and Matlab scripts |
tests | SVL regression tests |
Notes:
¹ Indicates that the directory contents are
not part of official SVL releases. The directories are under
still source control at Stanford, but not available publicly.
² Indicates that directory contents are excluded from source
control. Some scripts might assume that these directories exist.
SVL Libraries
-
The svlBase library provides base classes on which other library classes are built, or stand-alone abstract data types and utility classes.
-
The svlCuda library contains implementations of classes that are used for GPU acceleration of vision algorithms.
-
The svlDeprecated library contains classes that are no longer supported and will be removed in future releases.
-
The svlDevel library contains development (unstable) code that is not yet part of the main libraries. The code is generally buggier than the other libraries and the interface is subject to change between releases. Code from here is migrated to the other libraries once it becomes stable.
-
The svlML library provides basic machine learning capability. Some of the classes utilize the OpenCV machine learning library code.
-
The svlPGM library provides routines for inference and learning in Probabilistic Graphical Models (such as Bayesian Networks and Markov Random Fields).
-
The svlVision library implements 2d and 3d vision related functionality on top of OpenCV.
SVL Applications
[coming soon]
SVL Scripts
[coming soon]
Copyright © 2007-2009, Stephen Gould.