Group Actions
Engagement
  • Author Author: amgalbu
  • Date Created: Date Created
  • Views 475 views
  • Likes 0 likes
  • Comments 5 comments
Related
Recommended

Cam-e-lot - Round 3 - Man vs OpenCV

amgalbu

Hello everybody

For my project I need to make some analysis on the images captured by RaspiCam, so the next step is to install the OpenCV libraries and try to get images from RaspiCam inside a C program.

So here are the steps for this round (this tutorial has been adapted from the tutorial published by Pierre on Think RPI | Creative Raspberry Pi Area)

 

1. Install RaspiCam. Installation procedure is very well described on Raspberry official site http://www.raspberrypi.org/archives/3890 Once installed, test it with this command:

raspistill -t 10000

2. Download the MMAL library aspivid/Raspistill source code from https://github.com/raspberrypi/userland .

3. Unzip the file and copy the directory userlan-master in /opt/vc. I also renamed the directory as "userland"

4.  Go to /opt/vc/userland and type

     sed -i ‘s/if (DEFINED CMAKE_TOOLCHAIN_FILE)/if (NOT DEFINED CMAKE_TOOLCHAIN_FILE)/g’ makefiles/cmake/arm-linux.cmake

5. create a build directory...

     sudo mkdir build

     cd build

6. ... and compile

     sudo cmake -DCMAKE_BUILD_TYPE=Release ..
     sudo make
     sudo make install

7. go to /opt/vc/bin and test one file typing : ./raspistill -t 3000



OK... now I'm ready to create a new project

 

1. create a new folder (/home/pi/vc) in your home directory

     mkdir cv
     cd cv

2. copy all raspicam apps source code

     cp /opt/vc/userland/host_applications/linux/apps/raspicam/*  .

     mv RaspiStill.c camcv.c

     sudo chmod 777 camcv.c

3. remove then content of CMakeLists.txt and replace with :
     cmake_minimum_required(VERSION 2.8)
     project( camcv )
     SET(COMPILE_DEFINITIONS -Werror)
     include_directories(/opt/vc/userland/host_applications/linux/libs/bcm_host/include)
     include_directories(/opt/vc/userland/interface/vcos)
     include_directories(/opt/vc/userland)
     include_directories(/opt/vc/userland/interface/vcos/pthreads)
     include_directories(/opt/vc/userland/interface/vmcs_host/linux)
     include_directories(/opt/vc/userland/interface/khronos/include)

     include_directories(/opt/vc/userland/interface/khronos/common)

     add_executable(camcv RaspiCamControl.c RaspiCLI.c RaspiPreview.c camcv.c RaspiText.c RaspiTexUtil.c gl_scenes/teapot.c gl_scenes/models.c gl_scenes/square.c gl_scenes/mirror.c gl_scenes/yuv.c gl_scenes/sobel.c tga.c)
     target_link_libraries(camcv /opt/vc/lib/libmmal_core.so /opt/vc/lib/libmmal_util.so /opt/vc/lib/libmmal_vc_client.so /opt/vc/lib/libvcos.so /opt/vc/lib/libbcm_host.so /opt/vc/lib/libGLESv2.so /opt/vc/lib/libEGL.so)

4. delete CMakeFiles directory if it exits
5. compile & test
     cmake .
     make
     ./camcv -t 1000


Great!!!
We are now ready to install OpenCV


1. install both dev lib and python lib. Even if I'm going to use pure C for development, Python is still useful for small scripts, so I recommend to install it.

     sudo apt-get update

     sudo apt-get install libopencv-dev
     sudo apt-get install python-opencv

2. to test if OpenCv library is well installed, write this test software. It just displays a picture using imread and imshow functions. You will need to provide a sample .jpg file. To compile using OpenCv lib, create a CMakeLists.txt file with the following lines

     cmake_minimum_required(VERSION 2.8)
     project( displayimage )
     find_package( OpenCV REQUIRED )
     add_executable( displayimage display_image.cpp )
     target_link_libraries( displayimage ${OpenCV_LIBS} )

3. compile and execute

     cmake .
     make
     ./displayimage

4. Modify your CMakeFiles.txt to include OpenCV library

     project( camcv )

     SET(COMPILE_DEFINITIONS -Werror)

     #OPENCV
     find_package( OpenCV REQUIRED )

     include_directories(/opt/vc/userland/host_applications/linux/libs/bcm_host/include)

     include_directories(/opt/vc/userland/interface/vcos)
     include_directories(/opt/vc/userland)
     include_directories(/opt/vc/userland/interface/vcos/pthreads)
     include_directories(/opt/vc/userland/interface/vmcs_host/linux)
     include_directories(/opt/vc/userland/interface/khronos/include)

     include_directories(/opt/vc/userland/interface/khronos/common)

     add_executable(camcv RaspiCamControl.c RaspiCLI.c RaspiPreview.c camcv.c RaspiText.c RaspiTexUtil.c gl_scenes/teapot.c gl_scenes/models.c gl_scenes/square.c gl_scenes/mirror.c gl_scenes/yuv.c gl_scenes/sobel.c tga.c)
     target_link_libraries(camcv /opt/vc/lib/libmmal_core.so /opt/vc/lib/libmmal_util.so /opt/vc/lib/libmmal_vc_client.so /opt/vc/lib/libvcos.so /opt/vc/lib/libbcm_host.so /opt/vc/lib/libGLESv2.so /opt/vc/lib/libEGL.so ${OpenCV_LIBS})

5. Recompile. It should be ok, because we didn't change anything in the source code

     make
     ./camcv

 

I'm sorry this has been a very boring post with a little more than some shell scripting... In next post I will make a basic application that can capture an image from RaspiCam and make some video processing using the power of the OpenCV library

 

Stay tuned!

Parents

  • I've tried searching for this before, but could OpenCV be able to do facial recognition on cats ? image

     

    Frederick

  • in reply to fvan

    I'm not sure OpenCV has built-in face recognition features. For sure, this library https://github.com/bytefish/libfacerec works on Raspberry Pi and OpenCV. Probably with a proper training set it could be adapted to cat recognition..

  • in reply to amgalbu

    Hi,

    From openCV 2.4, this library is integrated into openCV so you will be able to use it without any extra installations. I worked with LBPH faceRecognizer and worked well for my project. But that time, it was human faces. In theory, it should work with cat faces also. But detecting cats faces may be a bit tricky. If face detector is not working, you could try eyes detector( which should most certainly work ) and then draw a rough estimation of face based on the position and size of detected eyes.

    You must first detect the faces in the frame and then pass it to the FaceRecognizer class. This may be a time consuming in Pi. My suggestion would be resize the captured frame to smaller size( I used 160x120 for mine ), detect the face, now you will be having the co-ordinates w.r.t the resized image, pass that area to the face recognizer and check the performance. If it's not working, do a little math and calculate the co-ordinates in the original frame and pass that area to your face recognizer. It's a lot more efficient when all you have is a single core system.

    Do share with us what results you are getting. image

    Cheers,

    vish

Comment
  • in reply to amgalbu

    Hi,

    From openCV 2.4, this library is integrated into openCV so you will be able to use it without any extra installations. I worked with LBPH faceRecognizer and worked well for my project. But that time, it was human faces. In theory, it should work with cat faces also. But detecting cats faces may be a bit tricky. If face detector is not working, you could try eyes detector( which should most certainly work ) and then draw a rough estimation of face based on the position and size of detected eyes.

    You must first detect the faces in the frame and then pass it to the FaceRecognizer class. This may be a time consuming in Pi. My suggestion would be resize the captured frame to smaller size( I used 160x120 for mine ), detect the face, now you will be having the co-ordinates w.r.t the resized image, pass that area to the face recognizer and check the performance. If it's not working, do a little math and calculate the co-ordinates in the original frame and pass that area to your face recognizer. It's a lot more efficient when all you have is a single core system.

    Do share with us what results you are getting. image

    Cheers,

    vish

Children
No Data