Title: COMS W4731 Final Project
Author: Varun Ravishankar
Dan Mercado
Email: vr2263@columbia.edu
dm2497@columbia.edu
Date: May 11th, 2013

This program tracks suspicious behavior in videos in real time.

This program was tested on OS X 10.8.3 and Ubuntu 12.04 LTS using Python 2.7. The program may run on older versions of Python; however, these are untested and unsupported.

Dependency versions that have been tested are listed here. Older versions of Python modules such as numpy or OpenCV may work as well; however, these are untested and unsupported.

• Python 2.7 or greater
• numpy 1.7.1
• OpenCV 2.4.5.0 or greater
• ffmpeg support [optional]
• Nvidia GPU with CUDA 5 support [optional]
• gcc [optional]
• Intel Threading Building Blocks
• Boost.Python 1.53.0 [optional]

The optional dependencies allow for GPU support, but you must build the provided GPU module first. You must also enable GPU support in your OpenCV build.

First, clone the repository and add it to your PYTHONPATH. Then, copy the path to a video you would like to analyze and pass it as an argument to read_video.py.

Next, you can run the Python scripts.

See an example below that starts off in the cloned repo:

$ pip install -r requirements.txt
$ make
$ python -O read_video.py data/video.mp4