能否更新下你的库

[movingh...@gmail.com]

yaowei，

你好，很感谢你的这个库，目前在人脸关键点检测方面这个是效果最好的。

现在我试用时出了一些问题：

环境：

g++ (Ubuntu 4.8.4-2ubuntu1~14.04) 4.8.4

Copyright (C) 2013 Free Software Foundation, Inc.

opencv版本: 3.0

操作系统: ubuntu14.04,64bit;

操作：

cd asmlibrary/linux

(makefile第15行改为：g++ -pg -o fit vjfacedetect.o video_camera.o demo_fit.o  -Wl,-rpath=./ -    L./ -lasmlibrary)

make

报错：

$ make

g++  -I/usr/local/include/opencv -O2 -Wall -pg -g -MD  -c -o demo_fit.o ../src/demo_fit.cpp

../src/demo_fit.cpp: In function ‘int main(int, char**)’:

../src/demo_fit.cpp:88:6: warning: variable ‘camera_idx’ set but not used [-Wunused-but-set-variable]

  int camera_idx = 0;

      ^

g++  -I/usr/local/include/opencv -O2 -Wall -pg -g -MD  -c -o vjfacedetect.o ../src/vjfacedetect.cpp

g++  -I/usr/local/include/opencv -O2 -Wall -pg -g -MD  -c -o video_camera.o ../src/video_camera.cpp

g++ -pg -o fit vjfacedetect.o video_camera.o demo_fit.o  -Wl,-rpath=./ -L./ -lasmlibrary 

/usr/bin/ld: skipping incompatible .//libasmlibrary.so when searching for -lasmlibrary

/usr/bin/ld: cannot find -lasmlibrary

collect2: error: ld returned 1 exit status

make: *** [fit] Error 1

猜测原因：

opencv的版本太久或者g++的版本太久；

请求：

能否更(基于opencv3.0)新下你的这个库

谢谢！

[Yao Wei]

64位系统应该用libasmlibrary64. so

Yao Wei
From Android's Gmail

--

---
You received this message because you are subscribed to the Google Groups "ASMLibrary" group.
To unsubscribe from this group and stop receiving emails from it, send an email to asmlibrary+...@googlegroups.com.
For more options, visit https://groups.google.com/d/optout.

[movingh...@gmail.com]

感谢你的回复！

后来把makefile文件中的15行改为：

g++ -pg -o fit vjfacedetect.o video_camera.o demo_fit.o  -Wl,-rpath=./ -    L./ -lasmlibrary64

编译成功，但是运行（已拷贝相应文件到此目录下）时报错：

$ ./fit -m my68-1d.amf -h haarcascade_frontalface_alt2.xml -i 9.jpg

 

/**************************************************************/

/*      ASMLibrary -- A compact SDK for face alignment        */

/*  Copyright (c) 2008-2011 by Yao Wei, all rights reserved.  */

/*                    Contact: nju...@gmail.com              */

/**************************************************************/

 

Reading active shape models from file my68-1d.amf ...(Done)

ASM model file read time cost: 24.27 millisec

Opencv haar-like file read time cost: 18.02 millisec

Segmentation fault (core dumped)

后来我跟踪错误：

(gdb) backtrace

#0  0x00007f852eca0b51 in cv::_InputArray::type(int) const () from ./libasmlibrary64.so

#1  0x00007f852b629315 in cv::Canny(cv::_InputArray const&, cv::_OutputArray const&, double, double, int, bool) ()

   from /usr/local/lib/libopencv_imgproc.so.3.0

#2  0x00007f852b62bd3c in cvCanny () from /usr/local/lib/libopencv_imgproc.so.3.0

#3  0x00007f852e94339c in cvHaarDetectObjectsForROC(void const*, CvHaarClassifierCascade*, CvMemStorage*, std::vector<int, std::allocator<int> >&, std::vector<double, std::allocator<double> >&, double, int, int, CvSize, CvSize, bool) () from /usr/local/lib/libopencv_objdetect.so.3.0

#4  0x00007f852e943d2b in cvHaarDetectObjects () from /usr/local/lib/libopencv_objdetect.so.3.0

#5  0x000000000040228d in detect_all_faces (Shapes=Shapes@entry=0x7fff5ee888d8, n_shapes=@0x7fff5ee888cc: 32645, image=0x1a7b810)

    at ../src/vjfacedetect.cpp:72

#6  0x0000000000401bed in main (argc=7, argv=<optimized out>) at ../src/demo_fit.cpp:236

再详细跟踪：

(gdb)  backtrace  full 

#0  0x00007f852eca0b51 in cv::_InputArray::type(int) const () from ./libasmlibrary64.so

No symbol table info available.

#1  0x00007f852b629315 in cv::Canny(cv::_InputArray const&, cv::_OutputArray const&, double, double, int, bool) ()

   from /usr/local/lib/libopencv_imgproc.so.3.0

No symbol table info available.

#2  0x00007f852b62bd3c in cvCanny () from /usr/local/lib/libopencv_imgproc.so.3.0

No symbol table info available.

#3  0x00007f852e94339c in cvHaarDetectObjectsForROC(void const*, CvHaarClassifierCascade*, CvMemStorage*, std::vector<int, std::allocator<int> >&, std::vector<double, std::allocator<double> >&, double, int, int, CvSize, CvSize, bool) () from /usr/local/lib/libopencv_objdetect.so.3.0

No symbol table info available.

#4  0x00007f852e943d2b in cvHaarDetectObjects () from /usr/local/lib/libopencv_objdetect.so.3.0

No symbol table info available.

#5  0x000000000040228d in detect_all_faces (Shapes=Shapes@entry=0x7fff5ee888d8, n_shapes=@0x7fff5ee888cc: 32645, image=0x1a7b810)

    at ../src/vjfacedetect.cpp:72

        pWork = 0x1a33ad0

        pFaces = <optimized out>

#6  0x0000000000401bed in main (argc=7, argv=<optimized out>) at ../src/demo_fit.cpp:236

        image = 0x1a7b810

        nFaces = 32645

        shapes = 0x0

        detshapes = 0x0

        flag = <optimized out>

        t = <optimized out>

        fit_asm = {m_model = {m_M = 0x197a9a0, m_B = 0x197b010, m_V = 0x197ae70, m_SM = 0x1984120, m_SSD = 0x19842c0, m_type = PROFILE_1D, {

              lbp_tdm = 0x1984450, classical_tdm = 0x1984450}, m_asm_meanshape = {m_vPoints = 0x197a760, m_nPoints = 68}, m_nPoints = 68, 

            m_nWidth = 17, m_nLevels = 4, m_nModes = 34, m_dReferenceFaceWidth = 167.935, m_bInterpolate = false, m_dMeanCost = 1149.04, 

            m_dVarCost = 210.226, m_CBackproject = 0x1a32470, m_CBs = 0x1a32940, m_dist = 0x1a32230, m_profile = 0x193bca0, m_search_shape = {

              m_vPoints = 0x1a32d00, m_nPoints = 68}, m_temp_shape = {m_vPoints = 0x1a32ad0, m_nPoints = 68}}, m_edge_start = 0x1a33440, 

          m_edge_end = 0x1a33160, m_nEdge = 180, m__VJdetavshape = {m_vPoints = 0x1a32f30, m_nPoints = 68}, m_param = {left = 1, right = 13}, 

          m_flag = false, m_dReferenceFaceWidth = 6.953222278231463e-310, __lastframe = 0x0, __pyrimg1 = 0x0, __pyrimg2 = 0x0, __features1 = 0x197a520, 

          __features2 = 0x1a33720, __found_feature = 0x1a33950 "", __feature_error = 0x1a339a0}

        model_name = <optimized out>

        cascade_name = <optimized out>

        n_iteration = <optimized out>

        t = <optimized out>

        filename = <optimized out>

        use_camera = <optimized out>

        image_or_video = <optimized out>

        camera_idx = <optimized out>

        i = <optimized out>

在 2016年2月17日星期三 UTC+8下午9:43:15，Yao Wei写道：

[Yao Wei]

这个是人脸检测部分出现问题，相关代码都在vjfacedetect. cpp中，不涉及到libasmlibrary。可以进一步调试。

Yao Wei
From Android's Gmail