Project update‎ > ‎

20121225-gen- Face pose estimation code study

posted Dec 25, 2012, 2:21 AM by 陳建儒
Face pose estimation所釋放的source code叫"PartsBasedDetector",在linux平台編譯使用,其主要要注意的事項有兩項,
  1. 使用那個函式偵測影像中的人臉
  2. 偵測到的人臉,其方向資訊儲存在那個變數
    其src/demo.cpp為detecting的主程式,裡面使用抓人臉函式為PartsBasedDetector為class寫在 PartsBasedDetector.cpp裡有寫到"brief search an image for potential object candidates",功能應為抓出可能為人臉的候選區域
    該函式做完後看似會有多個可能為人臉的候選區域,接著demo.cpp中有寫到用Visualize的class顯示最佳候選區域(display the best candidates),看似有從所有候選區域中識別出那些是人臉

接著該source code需用cmake 2.8.5 編譯用sudo apt-get install cmake指令只會更新到2.8.0,所以需自已上官網抓source code回來自行編譯安裝。

  1. 需要再安裝一個boost的函式庫,並指定路徑給cmak
    1. Installation and configuration的參考網頁"Getting Started on Unix Variants 
    2. Getting Started on Windows
  1. OpenCV路徑指向include,lib或source code都出現下列錯誤訊息:
    1. CMake Warning at CMakeLists.txt:86 (find_package):
    2.    By not providing "FindOpenCV.cmake" in CMAKE_MODULE_PATH this project has
    3.    asked CMake to find a package configuration file provided by "OpenCV", but
    4.    CMake did not find one.

    5.    Could not find a package configuration file provided by "OpenCV" with any
    6.    of the following names:

    7.      OpenCVConfig.cmake
    8.      opencv-config.cmake

    9.    Add the installation prefix of "OpenCV" to CMAKE_PREFIX_PATH or set
    10.    "OpenCV_DIR" to a directory containing one of the above files.  If "OpenCV"
    11.    provides a separate development package or SDK, be sure it has been
    12.    installed.