写在前面:前前后后使用了深度学习的TensorFlow(Keras)框架,Pytorch框架,最后是Caffe框架。三个框架各有特点,总的来说,Pytorch最易懂,面向对象编程,边运行边解释型,便于调试;TensorFlow次之,更倾向于面向过程编程,先设计静态图,使用Session运行Tensor,先解释再运行,对硬件,尤其是GPU的依赖更少,结合Keras高级API,可方便移植到移动端;Caffe是难度最大,同时也可能是性能最佳的,各种硬件环境都是提前编译,一次编译即可,缺点可能就是调试不方便。
入坑Caffe配置的过程笔记 *\ ^ _ ^ /*。
硬件环境:
Ubuntu GPU:CUDA9.0
OpenCV版本:OpenCV-3.4.1
OpenCV-3.4.1下载网址:OpenCV-3.4.1 提取码:9pyl
1、安装环境依赖:
sudo apt-get install build-essential
sudo apt-get install cmake git libgtk2.0-dev pkg-config libavcodec-dev libavformat-dev libswscale-dev
sudo apt-get install python-dev python-numpy libtbb2 libtbb-dev
sudo apt-get install libjpeg-dev libpng-dev libtiff5-dev libdc1394-22-dev # 处理图像所需的包
sudo apt-get install libavcodec-dev libavformat-dev libswscale-dev libv4l-dev liblapacke-dev
sudo apt-get install libxvidcore-dev libx264-dev # 处理视频所需的包
sudo apt-get install libatlas-base-dev gfortran # 优化opencv功能
2、编译OpenCV
cd opencv-3.4.1
mkdir mybuild
cd mybuild
cmake -D CMAKE_BUILD_TYPE=Release -D CMAKE_INSTALL_PREFIX=/usr/local ..
make
sudo make install
3、配置bash:
sudo /bin/bash -c 'echo "/usr/local/lib" > /etc/ld.so.conf.d/opencv.conf'
sudo source /etc/bash.bashrc
#add the following to bash.bashrc
PKG_CONFIG_PATH=$PKG_CONFIG_PATH:/usr/local/lib/pkgconfig 4 export PKG_CONFIG_PATH
#setup linkconfig
sudo ldconfig
#activate bash.bashrc
sudo source /etc/bash.bashrc
4、查看是否安装成功:
pkg-config --modversion opencv
可能出现的问题:
1、编译时出现:
-- Configuring incomplete, errors occurred
解决办法:
进入mybuild文件,删除CMakeCache.txt,重新编译:
cmake -D CMAKE_BUILD_TYPE=Release -D CMAKE_INSTALL_PREFIX=/usr/local ..