This is a collection of Linux applications based on OpenCV. The development machine is Ubuntu Linux 14.04, but it should also work in Ubuntu Linux 12.04 LTS. The OpenCV version of development is OpenCV 2.4.9 (2014-04-25). To summarize, the best runtime environment, which is tested, will be as follows.
- Operating System: Ubuntu Linux 14.04
- OpenCV Version: OpenCV 2.4.9 (2014-04-25)
In this section, the installation procedure will be discussed.
This collection of applications require OpenCV libraries. To install OpenCV, you can follow these steps.
-
Download OpenCV source code of version 2.4.9 for Linux from
http://opencv.org/downloads.html -
Extract the downloaded package to your workspace. Notice that these files must NOT be removed after installation, because OpenCV application require the libraries from the installation path. So you should find a good place for the extracted files.
-
Run the following command at the root path of the extracted package to generate the makefile.
$ cmake . -
Then run the following command to compile the OpenCV source codes.
$ make -
After compiling the source code, run the following command to install the OpenCV. This command require administrator power.
$ sudo make install
If there is no error, then you have successfully installed the OpenCV on your machine.
This project only requires OpenCV and Linux environments as the runtime. After that, you just need to copy the entire project to your workspace.
The compiling instructions will be discussed in the next section.
In this section, the compiling and executing procedures of the involved applications will be discussed.
All applications can be compiled to an executable program through
the same way. The generate.sh script at the root path is
provided as an easy tool to help compile the source codes to the
corresponding executable programs.
One source file in the src/ directory represents an application
and you can use the following commnad to query all applications
available to be generated into executable program.
./generate.sh
After running the above command, the usage of this script in blue color and names of all available applications in yellow color will be printed on the screen.
Notice that the script should have execution property. If it does not, use the following command to make it executable.
sudo chmod 777 generate.sh
The above command gives the execution property to the script.
Use the following command to generate, or says, compile an application.
./generate.sh <app>
This argument <app> is the name of the application to compile.
All available applications can be found via the method mentioned
in the previous section.
Notice that the name of application does NOT include its file name extension.
For example, if the application to compile is Webcam, then run
the following command.
./generate.sh Webcam
The letter case should be considered, and after running the above
command, the corresponding executable program Webcam.out will be
generated in the release/ directory.
The usages of different applications will be discussed in this section. Yet some applications have usage discription built internally and those information will be shown if you have used them in wrong ways.
This application shows how to display load an image from file and display it with OpenCV. The usage is
./DisplayImage.out <image_path>
Where the argument <image_path> is the file path of the image
file to display.
Webcam video stream displaying demo with OpenCV. The usage is as follows.
./Webcam.out <device_id>
Where the argument <device_id> is the video device identifier to
use. Usually it is 0 if only one webcam connects to the machine.
This application perform face detection from an image file. The detection result will be displayed on the screen as an processed image. The usage is as follows.
./FaceDetectionImage.out <cascade_name> <image_name>
Where
-
<cascade_name>is the path of the pre-trained Haar-cascade of for face detection. -
<image_name>is the path of the image file to detect face from.
This application helps collect facial data for face recognition. Its usage is as follows.
./FaceCollection.out <cascade_name> <device_id>
Where
-
<cascade_name>is the path of the pre-trained Haar-cascade of for face detection. -
<device_id>is the video device identifier to capture video stream from. Usually it is0if only one webcam connects to the machine.
This application combines face detection and image corpping
functions. A rectangle will occur on the video if any face is
detected. And the user just need to press [space] on the
keyboard to save the face image as an image file.
After collecting face images, you can place them in corresponding
directory structures and use the create_csv.sh script file to
generate the CSV (Comma Separated Value) file. This file tags face
images with corresponding identification number and subject name.
The standard directory structure can be found in the discription
within the source code of the script.
This application performs face recognition from video stream using OpenCV. Its usage is as follows.
./FaceRecognition.out <cascade> <csv> <device_id>
Where
-
<cascade>is the path of the pre-trained Haar-cascade of for face detection. -
<csv>is the path of the CSV file that represents the face database to refer to. -
<device_id>is the video device identifier to capture video stream from. Usually it is0if only one webcam connects to the machine.
This application is a combination of face recognition and face collection from video stream using OpenCV. Below is the usage.
./FaceRecognition2.out <cascade> <csv> <device_id>
Where
-
<cascade>is the path of the pre-trained Haar-cascade of for face detection. -
<csv>is the path of the CSV file that represents the face database to refer to. -
<device_id>is the video device identifier to capture video stream from. Usually it is0if only one webcam connects to the machine.
A rectangle with the predicted name of the subject will occur on
the video if any face is detected. And if the prediction does not
match actual user, one just need to press [space] on keyboard to
save the current face image as an image file. After placing these
collected images in correct directories and use the create_csv.sh
script file to generate the CSV file, the new face data are pushed
to the face database.
This application performs face recognition from image file using OpenCV. The usage is as follows.
./FaceRecognitionImage.out <cascade> <csv> <in_image> <out_info> [<out_image>]
Where
-
<cascade_name>is the path of the pre-trained Haar-cascade of for face detection. -
<csv>is the path of the CSV file that represents the face database to refer to. -
<in_image>is the path of the image file to perform face recognition on. -
<out_info>is the expected path to put the face recognition result. The output result is in JSON format. -
<out_image>is the expected path to put the output image file after face recognition. A rectangle and corresponding tags will occur if any face is detected. This argument is optional, and if it is not provided, no output image will be generated.
Below is the directory structure and corresponding introduction.
-
src/: Source code directory. All the source codes of the applications will be placed in this directory. Each souce code file in this directory represent one application, or says, one executable program. -
build/: Build directory. This directory includes all files of compiling process. During compiling, temporary files generated will be placed here../CMakeLists.txt.template: CMakeLists template. It has the commands for cmake, but it cannot be used directly. This template file will be used by thegenerate.shscript to create correspondingCMakeLists.txtfile for different applications.
-
release/: Executable application release directory. The compiled executable applications will be placed here after they are compiled. Some data files will also be placed here, but those files are not a part of the release applications. -
generate.sh: Application generation script. This script will help generate different applications. -
README.md: Read me document. It is this document, which gives introduction to the project and some instructions.
Below is the copyright announcement and the contact of the author.
- Author: David Qiu
- Email: david@davidqiu.com
- Email: dicong.qiu@intel.com
- Website: http://www.davidqiu.com/
Copyright (C) 2014, David Qiu. All rights reserved.
Notice that there are some data files in the release directory
and these files are eithor published on the Internet or collected
by the author of this project.
For those published data files, the copyright belong to the original authors.
For those data files collected by the author of this project, the copyright belongs to the author of this project, and must NOT be used in any circumstance without approvement from the author of this project.