Vision-Based System for Automated Mouse Behavior Recognition - System requirements and Installation instructions

Installation instructions

Tested Configuration:

The install scripts are designed specifically to simplify installation for this specific configuration. Different configurations will generally require changes to install paths and required packages.

Installation Steps:

  1. Download and unpack the software
  2. cd into the software root and run the "install" script. The script will ask for your root password in order to install required dependencies. It will then compile the required dependencies on your system.
  3. GPU performance improvement (recommended) - skip this step if you do not want the massive GPU speedup
    1. Download and install the NVIDIA CUDA toolkit and the CUDA driver, version 3.2. Make sure to install the driver first.
    2. Run the "installCns" script. Close Matlab when the script completes.
  4. Run the "start_system" script. This script sets the necessary environment variables, and then starts matlab with the mouse system GUI.

Content:

  • Representation/cns: framework written by Jim Mutch & Ulf Knoblich for the fast simulation of cortically-organized networks (click here for instllation guide)
  • Representation/hjpkg: a CNS package for computing motion features.
  • Classification/: contains code to do the classification for the trainign system
  • gui: contains code used in the gui and code to allow the gui to interact with the backend training and testing system
  • imageprocessing: code to process images (crop, invert, etc.)
  • tracking: code to track the mouse
  • training: code to train a system on new behaviors
  • videoprocessing: code to process videos
  • visualize: code to save testing results to a video file
  • third_party/opencv-2.1.0: an image processing library, for reading/saving compressed videos and computing background (Downloaded at http://sourceforge.net/projects/opencvlibrary/)
  • third_party/SVMHMM: a classifier written by Thorsten Joachims for classifying video sequences (Downloaded at http://www.cs.cornell.edu/People/tj/svm_light/svm_hmm.html)