iLab Neuromorphic Vision Toolkit Documentation

Version 3.1

Welcome to the iLab Neuromorphic Vision C++ Toolkit. This toolkit is being developed at USC and Caltech to provide a comprehensive suite of C++ classes for image processing, biologically-inspired computational vision, hardware interfacing, and parallel processing on Linux beowulf clusters. Among the executables included in the distribution are our bottom-up, saliency-based visual attention model (single-CPU and beowulf versions), models of contour integration in primary visual cortex, models of object recognition in inferotemporal cortex, and our beobots quad-CPU autonomous neuromorphic robot code. Please see the README for installation and compilation instructions.

The main program to use is called ezvision in the saliency/bin/ directory.

See the README and programmer notes for detailed information on building and using the programs in the toolkit.

Here are links to several key classes in the toolkit:

• Image
• ModelComponent
• ModelManager
• ChannelBase
• VisualCortex
• Brain
• InputFrameSeries
• OutputFrameSeries

Other pages of interest:

• building models using components of this toolkit
• Introduction to the C++/Tcl scripting engine

To further help you navigate the source tree, see the available viewcvs browser (use the same username and password as you used to access this page) which, despite the "cvs" in its name, browses our subversion (svn) source code repository.