Skip to content
This repository was archived by the owner on Jan 27, 2024. It is now read-only.

Home

Jump to bottom
kovbal edited this page May 30, 2015 · 8 revisions

FaceGuard

This wiki contains the documentation for our coursework of the biometric identification course on the University of Pannonia.

Our goal was to identify people using a webcam based on a small set of training picture.

The application is written in C++ using the Qt framework and the OpenCV library. It was tested under Windows 7, Windows 8.1 and Fedora 19.

The purpose of this wiki is to document our application in order to easen further development. We intend to provide a high-level overview of the software for developers. Basic knowledge of C++ programming, OpenCV, and Qt is assumed. It only serves as an introduction, and does not describe every detail of the application—that knowledge is only obtainable by reading the source code. It is probably advisable to start the reading with the GUI page.

Basic assumptions

During the develeopment of our application we assumed that the user is cooperative.

As stated earlier, our goal was to identify people, i.e. decide which person from a database sits in front of the computer.

How to compile

In order to compile the application, the Qt framework and the OpenCV library must be installed.

Under Windows the OPENCV_PATH and the OPENCV_VERSION qmake variable must be set correctly during the Makefile generation. These values can be set in Qt Creator under the label Additional arguments (Projects → Build Settings → Build Steps → qmake → Additional arguments). Example values:

  • OPENCV_PATH=../../opencv
  • OPENCV_VERSION=2411

Under Linux, the OpenCV library is often placed somewhere in the PATH, therefore the above parameters does not need to be set. However, if it is located elsewhere then the variables can be set accordingly.

After compilation, the following files must be placed in the executable's directory:

  • haarcascade_frontalface_default.xml
  • haarcascade_eye.xml
  • haarcascade_mcs_eyepair_small.xml
  • haarcascade_mcs_lefteye.xml
  • haarcascade_mcs_righteye.xml

The generated Makefile's install target copies all neccessary files to the executable's directory, including the aforementioned XMLs and the required Qt and OpenCV DLLs. The easiest way to run the install step is to add a "custom build step" to Qt Creator (Projects → Build Steps → Add Build Step → "Make arguments": install).

Deployment under Windows

After the generated Makefile's install target has ran, the only files missing are the DLLs for the C++ runtime as stated in the Qt deployment guide. In order to deploy the application on Windows one must copy these files to the target machine.

License

The application is licensed under the BSD license.

Clone this wiki locally