代码程序相关问题故障解决目录

1. Python

1.1 OpenCV

1.1.1 由于OpenCV接口更改导致的错误

• error: ‘CV_LOAD_IMAGE_UNCHANGED’ was not declared in this scope
  CV_LOAD_IMAGE_UNCHANGED改为cv::IMREAD_UNCHANGED

• error: ‘CV_LOAD_IMAGE_GRAYSCALE’ was not declared in this scope
  CV_LOAD_IMAGE_GRAYSCALE改为IMREAD_GRAYSCALE

• error: ‘CV_MINMAX’ was not declared in this scope
  CV_MINMAX改成NORM_MINMAX

• error: ‘CV_BGR2GRAY’ was not declared in this scope cvtColor
  添加#include <opencv2\imgproc\types_c.h>

1.2 Tensorflow

1.2.1 模型训练

• Process finished with exit code 137 (interrupted by signal 9: SIGKILL)

  原因：

  程序并没有运行完，系统强制杀死进程，一般来说是因为内存不足或者CPU不够用。
  内存扩张可能原因：loss或者网络的输出不断积累导致计算图不断扩张

  解决办法：

  1. 选择小的batch_size，或者不要把一整个大的文件一次性读到内存里；
  2. 在训练的循环过程中，需要用到loss，则用loss.data[0]

  参考：

  Process finished with exit code 137 (interrupted by signal 9: SIGKILL)
  tensorflow: interrupted by signal 9: SIGKILL
  Pytorch Process finished with exit code137(interrupted by signal9:SIGKILL)

---

2. C++

3. 其他

• 「git失败」gnutls_handshake() failed: The TLS connection was non-properly terminated.
  重置代理

  git config --global  --unset https.https://github.com.proxy 
  git config --global  --unset http.https://github.com.proxy 
  

• Failed to load module “canberra-gtk-module”

  GTK+ 2.x symbols detected. Using GTK+ 2.x and GTK+ 3 in the same process is not supported.
  Failed to load module “canberra-gtk-module”

• 解决 canberra-gtk-module 加载失败的问题

  sudo apt-get install libcanberra-gtk-module -y
  sudo ln -s /usr/lib/x86_64-linux-gnu/gtk-2.0/modules/libcanberra-gtk-module.so /usr/lib/libcanberra-gtk-module.so
  

• relocation R_X86_64_PC32 against symbol `XXXX’ can not be used when making a shared object; recompile with -fPIC
  有冲突，如在安装opencv时，与gflags冲突：

  [ 94%] Linking CXX shared library …/…/lib/libopencv_sfm.so
  /usr/bin/ld: /usr/local/lib/libgflags_nothreads.a(gflags.cc.o): relocation R_X86_64_PC32 against symbol `_ZN22gflags_mutex_namespace5MutexD1Ev’ can not be used when making a shared object; recompile with -fPIC
  /usr/bin/ld: 最后的链结失败: 错误的值
  collect2: error: ld returned 1 exit status
  modules/sfm/CMakeFiles/opencv_sfm.dir/build.make:327: recipe for target ‘lib/libopencv_sfm.so.4.5.2’ failed

  可将libgflags_nothreads.a重命名再安装，或先卸载gflags，或交换gflags与opencv的安装次序。

• 明明已经安装了某个模块，python报错：ImportError: No module named ***

• [glog安装]Could NOT find GTest (missing: GTEST_LIBRARY GTEST_MAIN_LIBRARY)

  GTest虽然安装过，但是找不到头文件之类
  解决方法如下：
  sudo apt-get install libgtest-dev
  sudo apt-get install cmake
  cd
  cd /usr/src/gtest
  sudo cmake CMakeLists.txt
  sudo make
  sudo cp *.a /usr/lib
  原文链接：https://blog.csdn.net/Numberors/article/details/109765131

• [glog安装]Could NOT find Unwind (missing: Unwind_INCLUDE_DIR Unwind_LIBRARY Unwind_PLATFORM_LIBRARY)

  sudo apt-get install libunwind-dev
  

• [opencv安装，进行sudo make时]make[2]: *** 没有规则可制作目标“/usr/local/lib/libgflags_nothreads.a”，由“lib/libopencv_sfm.so.4.5.2” 需求。 停止。
  在usr/lib下查找libgflags_nothreads.a，将其软连接到/usr/local/lib/libgflags_nothreads.a，解决。

• 使用matplotlib显示中文字符 －> 解决方法

• Failed to get convolution algorithm. This is probably because cuDNN failed to initialize －> 解决方法