配置

find:
cannot find -lcudart
解决：
sudo ln -s /usr/local/cuda/lib64/libcudart.so /usr/lib/libcudart.so

1: TensorRT7.1.3.4安装

下载链接：https://developer.nvidia.com/tensorrt

https://docs.nvidia.com/deeplearning/tensorrt/install-guide/index.html#installing

tar -xvzf TensorRT-7.1.3.4.Ubuntu-16.04.x86_64-gnu.cuda-10.2.cudnn8.0.tar.gz

cd data
python3 ./download_pgms.py
cd sample
pip install pillow
../bin/sample_mnist

export LD_LIBRARY_PATH=/home/cs/TensorRT-7.1.3.4/lib:$LD_LIBRARY_PATH


cd /TensorRT-7.1.3.4/python
python3.6 -m pip install tensorrt-7.1.3.4-cp36-none-linux_x86_64.whl

将TensorRT中的链接文件.so文件复制到/usr/lib/文件夹中，比如

1） ImportError: libnvinfer.so.6: cannot open shared object file:
Nosuch file or directory

2）ImportError: libnvonnxparser.so.6: cannot open shared object file:No
such file or directory

解决办法：

1） sudo cp TensorRT-6.01/targets/x86_64-linux-gnu/lib/libnvinfer.so.6
/usr/lib/

2）sudo cp
TensorRT-6.01/targets/x86_64-linux-gnu/lib/libnvonnxparser.so.6
/usr/lib/

出现pycuda问题
ModuleNotFoundError: No module named ‘pycuda’

sudo pip3.6 install pycuda== 2019.1.2 -i
https://pypi.tuna.tsinghua.edu.cn/simple
// 不用sudo
pip3.6 install pycuda -i https://pypi.tuna.tsinghua.edu.cn/simple
出现 is not a symbolic link
sbin/ldconfig.real: /usr/local/cuda-11.1/targets/x86_64-linux/lib/libcudnn_ops_
infer.so.8 is not a symbolic link
sudo ln -sf libcudnn.so.8.0.2 libcudnn.so.8

出现

CMakeFiles/traffic_det_reg_caffe_trt.dir/src/TrafficDetection.cpp.o:
In function onnxToTRTModel(std::__cxx11::basic_string<char, std::char_traits<char>, std::allocator<char> > const&, std::__cxx11::basic_string<char, std::char_traits<char>, std::allocator<char> > const&, nvinfer1::ICudaEngine*&, int const&)': TrafficDetection.cpp:(.text+0x1357): undefined reference tocreateNvOnnxParser_INTERNAL’

cmakelist添加：

/home/name/TensorRT-7.1.3.4/lib/libnvinfer.so
/home/name/TensorRT-7.1.3.4/lib/libnvinfer_plugin.so
/home/name/TensorRT-7.1.3.4/lib/libnvparsers.so
/home/name/TensorRT-7.1.3.4/lib/libnvonnxparser.so

2: TensorRT7.1.3.4 C++测试

ICudaEngine类即为Engine，可通过IBuilder类方法buildCudaEngine()/buildEngineWithConfig()返回其指针。
注意，可通过导入模型生成Engine和通过反序列化来加载Engine两种Engine生成方式。

 //method 1:
ICudaEngine *engine = builder-buildCudaEngine(*network);
 // method 2:
nvinfer1::IPluginFactory *mPlugin;
mEngine = shared_ptr<nvinfer1::ICudaEngine>(
	 	builder->buildCudaEngine(*network), InferDeleter());
//method 3:	
nvinfer1::IPluginFactory *mPlugin;
mEngine = SampleUniquePtr<nvinfer1::ICudaEngine>(
        builder->buildEngineWithConfig(*network, *config));

例如，输入输出通道的转换：
分割输入通道转换：

            cv::Mat img = cv::imread();
            if (img.empty()) continue;
            // BGR to RGB
            cv::Mat pr_img = preprocess_img(img); 
            int i = 0;
            for (int row = 0; row < INPUT_H; ++row) {
    
                uchar* uc_pixel = pr_img.data + row * pr_img.step;
                for (int col = 0; col < INPUT_W; ++col) {
    
                    data[b * 3 * INPUT_H * INPUT_W + i] = (float)uc_pixel[2] / 255.0;
                    data[b * 3 * INPUT_H * INPUT_W + i + INPUT_H * INPUT_W] = (float)uc_pixel[1] / 255.0;
                    data[b * 3 * INPUT_H * INPUT_W + i + 2 * INPUT_H * INPUT_W] = (float)uc_pixel[0] / 255.0;
                    uc_pixel += 3;
                    ++i;
                }
            }

分割输出通道转换：

//在这里	
        doInference(*context, data, prob, 1); //chw
		out = chw2hwc(prob, out);
		//hwc
		cv::resize(out, real_out, real_out.size());
		real_out_ = map2three(real_out, real_out_);

3: troch C++测试

// libtroch版本可能会引起ros未定义
set(Torch_DIR /home/libtorch/share/cmake/Torch) 
find_package(Torch REQUIRED)
find_package(Torch PATHS ${
    Torch_DIR} NO_DEFAULT REQUIRED)
if (Torch_FOUND)
    message(STATUS "Torch library found!")
    message(STATUS "    include path: ${TORCH_INCLUDE_DIRS}" \n)
else ()
    message(FATAL_ERROR "Could not locate Torch" \n)
endif()

选择版本下载
https://pytorch.org/

https://download.pytorch.org/libtorch/cu102/libtorch-shared-with-deps-1.7.1.zip

unzip libtorch-shared-with-deps-1.7.1+cu101.zip

解压后
实例代码

https://pytorch.org/cppdocs/installing.html

#include <torch/torch.h>
#include <iostream>

int main() {
    
  torch::Tensor tensor = torch::rand({
    2, 3});
  std::cout << tensor << std::endl;
}

cmake_minimum_required(VERSION 3.0 FATAL_ERROR)
project(example-app)

find_package(Torch REQUIRED)
set(CMAKE_CXX_FLAGS "${CMAKE_CXX_FLAGS} ${TORCH_CXX_FLAGS}")

add_executable(example-app example-app.cpp)
target_link_libraries(example-app "${TORCH_LIBRARIES}")
set_property(TARGET example-app PROPERTY CXX_STANDARD 14)

# The following code block is suggested to be used on Windows.
# According to https://github.com/pytorch/pytorch/issues/25457,
# the DLLs need to be copied to avoid memory errors.
if (MSVC)
  file(GLOB TORCH_DLLS "${TORCH_INSTALL_PREFIX}/lib/*.dll")
  add_custom_command(TARGET example-app
                     POST_BUILD
                     COMMAND ${
    CMAKE_COMMAND} -E copy_if_different
                     ${
    TORCH_DLLS}
                     $<TARGET_FILE_DIR:example-app>)
endif (MSVC)

mkdir example
cd example/
vim CMakeLists.txt
vim example-app.cpp
mkdir build
cd build/
cmake -DCMAKE_PREFIX_PATH=/home/libtorch ..
cmake --build . --config Release
./example-app

project(test_pytorch)
set(CMAKE_CXX_STANDARD 14)
set(Torch_DIR /home/zyt/libtorch/share/cmake/Torch) 
 SET(CMAKE_BUILD_TYPE "RELEASE") 
#include_directories(${OpenCV_INCLUDE_DIRS}
find_package(OpenCV REQUIRED)
find_package(Torch REQUIRED)

add_executable(test_pytorch test_pytorch.cpp)
target_link_libraries(test_pytorch ${
    OpenCV_LIBS}  ${
    TORCH_LIBRARIES}) 

mkdir build
cd build/
cmake ..
make
./test_pytorch

load .pt模型

import torch
import torchvision
from unet import UNet
model = UNet(3, 2)#自己定义的网络模型
model.load_state_dict(torch.load("weights.pth"))#保存的训练模型
model.eval()#切换到eval（）
example = torch.rand(1, 3, 320, 480)#生成一个随机输入维度的输入
traced_script_module = torch.jit.trace(model, example)
traced_script_module.save("model.pt")

Detector::Detector(const std::string& model_path, const torch::DeviceType& device_type) : device_(device_type) {
    
    try {
    
        module_ = torch::jit::load(model_path);
    }
    catch (const c10::Error& e) {
    
        std::cerr << "Error loading the model!\n";
        std::exit(EXIT_FAILURE);
    }
    half_ = (device_ != torch::kCPU);
    module_.to(device_);
    if (half_) {
    
        module_.to(torch::kHalf);
    }
    module_.eval();
}

infer

  t ensor_img = tensor_img.permute({
    0, 3, 1, 2}).contiguous();  // BHWC -> BCHW (Batch, Channel, Height, Width)
    if (half_) {
    
        tensor_img = tensor_img.to(torch::kHalf);
    }
    std::vector<torch::jit::IValue> inputs;
    inputs.emplace_back(tensor_img);
    /*** Inference ***/
    // inference
    torch::jit::IValue output = module_.forward(inputs);
   auto detections = output.toTuple()->elements()[0].toTensor();
   auto result = PostProcessing(detections, pad_w, pad_h, scale, img.size(), conf_threshold, iou_threshold);

3: ROS下配置TRT和libtorch测试

set(CMAKE_BUILD_TYPE "RELEASE") 
set(Torch_DIR /home/libtorch/share/cmake/Torch) 
find_package(Torch REQUIRED)
find_package(Torch PATHS ${
    Torch_DIR} NO_DEFAULT REQUIRED)
if (Torch_FOUND)
    message(STATUS "Torch library found!")
    message(STATUS "    include path: ${TORCH_INCLUDE_DIRS}" \n)
else ()
    message(FATAL_ERROR "Could not locate Torch" \n)
endif()

set(CMAKE_CXX_STANDARD 14)
set(CMAKE_CXX_FLAGS "${CMAKE_CXX_FLAGS} -std=c++11")
  catkin_package(
    CATKIN_DEPENDS
      roslib
      pcl_ros
  )
  include_directories(
    include
    ${
    catkin_INCLUDE_DIRS}
  )
INCLUDE_DIRECTORIES(
        /home/TensorRT-7.0.0.11/include 
        /home/TensorRT-7.0.0.11/samples/)

错误一：
catkin_make无法通过，一直在报错如下：

CMakeFiles/detect_node.dir/src/detect_node.cpp.o: In function main': detect_node.cpp:(.text+0x118): undefined reference to ros::init(int&, char**, std::string const&, unsigned int)'
detect_node.cpp:(.text+0x1aa): undefined reference to ros::NodeHandle::NodeHandle(std::string const&, std::map<std::string, std::string, std::less<std::string>, std::allocator<std::pair<std::string const, std::string> > > const&)' detect_node.cpp:(.text+0x28d): undefined reference to
。。。。。。。。。

一般：
这个问题一般都是在cmakelists.txt中没有加入相应的链接库导致的，但是明明加了！
libtorch版本的问题！可能最新版本的libtorch库跟ROS的库一起链接的时候会不兼容吧

错误二：

pcl::io::loadPCDFile<pcl::PointXYZI> //无法通过，
undefined reference to `pcl::PCDReader::read(std::string const&, pcl::PCLPointCloud2&, Eigen::Matrix<float, 4, 1, 0, 4, 1>&, Eigen::Quaternion<float, 0>&, int&, int)

一直在报错，可能也是libtorch版本导致的，换版本能解决
调整14 11
set(CMAKE_CXX_STANDARD 14)
set(CMAKE_CXX_FLAGS "${CMAKE_CXX_FLAGS} -std=c++11")

使用多线程
set(CMAKE_CXX_FLAGS "${CMAKE_CXX_FLAGS} -std=c++11  -pthread" )

3: ROS下配置caffe测试

1、在指定的目录下，下载需要编译的caffe
创建目录XXX，在该目录下右键打开终端，clone 需要编译的caffe

git clone https://github.com/BVLC/caffe.git

2、进入caffe目录，复制Makefile.config.example 到 Makefile.config 中，并编辑修改Makefile.config中内容
进入 caffe ，将 Makefile.config.example 文件复制一份并更名为 Makefile.config ，也可以在 caffe 目录下直接调用以下命令完成复制操作 ：

cd caffe
 
ls
 
sudo cp Makefile.config.example Makefile.config
 
sudo gedit Makefile.config

打开Makefile.config文件，进行如下修改：

1.应用 cudnn

将
#USE_CUDNN := 1
修改成：
USE_CUDNN := 1

2.应用 opencv 版本

将
#OPENCV_VERSION := 3
修改为：
OPENCV_VERSION := 3

3.使用 python 接口

将
#WITH_PYTHON_LAYER := 1
修改为
WITH_PYTHON_LAYER := 1

4.修改 python 路径

INCLUDE_DIRS := $(PYTHON_INCLUDE) /usr/local/include
LIBRARY_DIRS := $(PYTHON_LIB) /usr/local/lib /usr/lib
修改为：
INCLUDE_DIRS := $(PYTHON_INCLUDE) /usr/local/include /usr/include/hdf5/serial
LIBRARY_DIRS := $(PYTHON_LIB) /usr/local/lib /usr/lib /usr/lib/x86_64-linux-gnu /usr/lib/x86_64-linux-gnu/hdf5/serial  
  5.CUDA_RACH 。 因为我安装的cuda版本较高，compute_20和compute_21已经过时，所以注释掉这两行

将
CUDA_ARCH := -gencode arch=compute_20,code=sm_20 \
		-gencode arch=compute_20,code=sm_21 \
 
修改为：
CUDA_ARCH := #-gencode arch=compute_20,code=sm_20 \
		#-gencode arch=compute_20,code=sm_21 \
这里给出我修改后的Makefile.config文件内容，我的配置文件详细内容见文末：

3、然后修改 caffe 目录下的 Makefile 文件
sudo gedit Makefile
修改如下两项：

将：
NVCCFLAGS +=-ccbin=$(CXX) -Xcompiler-fPIC $(COMMON_FLAGS)
替换为：
NVCCFLAGS += -D_FORCE_INLINES -ccbin=$(CXX) -Xcompiler -fPIC $(COMMON_FLAGS)
    

将：
LIBRARIES += glog gflags protobuf boost_system boost_filesystem m hdf5_hl hdf5
改为：
LIBRARIES += glog gflags protobuf boost_system boost_filesystem m hdf5_serial_hl hdf5_serial

出现:

/usr/include/c++/5/bits/c++0x_warning.h:32:2: error: #error This file requires compiler and library support for the ISO C++ 2011 standard. This support must be enabled with the -std=c++11 or -std=gnu++11 compiler options.

解决方法

解决：修改Makefile文件

CXXFLAGS += -pthread -fPIC $(COMMON_FLAGS) $(WARNINGS) -std=c++11
NVCCFLAGS += -D_FORCE_INLINES -ccbin=$(CXX) -Xcompiler -fPIC $(COMMON_FLAGS) -std=c++11
LINKFLAGS += -pthread -fPIC $(COMMON_FLAGS) $(WARNINGS) -std=c++11

出现

fatal error: opencv2/core/core.hpp: No such file or directory

解决makefile添加
#LIBRARIES += glog gflags protobuf boost_system boost_filesystem m
LIBRARIES += glog gflags protobuf boost_system boost_filesystem m hdf5_serial_hl hdf5_serial opencv_core opencv_highgui opencv_imgproc opencv_imgcodecs

源码安装opencv
https://blog.csdn.net/fightingboom/article/details/88732537

unzip opencv-3.4.3
unzip opencv_contrib-3.4.3
sudo cmake -D CMAKE_BUILD_TYPE=RELEASE -D CMAKE_INSTALL_PREFIX=/usr/local -D OPENCV_EXTRA_MODULES_PATH=/home/opencv_contrib-3.4.3/modules -D INSTALL_PYTHON_EXAMPLES=ON -D BUILD_EXAMPLES=ON ..

sudo cmake   -D CMAKE_BUILD_TYPE=RELEASE   -D CMAKE_INSTALL_PREFIX=/usr/local   -D WITH_TBB=ON   -D WITH_V4L=ON -D WITH_LIBV4L=ON   -D WITH_CUDA=ON   -D ENABLE_FAST_MATH=1 -D OPENCV_ENABLE_NONFREE=ON  -D CUDA_FAST_MATH=1   -D CUDA_NVCC_FLAGS="-D_FORCE_INLINES"  -D WITH_CUBLAS=1 -D OPENCV_EXTRA_MODULES_PATH=../../opencv_contrib-3.4.3/modules -D BUILD_opencv_dnn=ON ..

sudo make -j4

sudo make install

sudo make

sudo make install

sudo ldconfig

sudo cmake -D CMAKE_BUILD_TYPE=RELEASE -D CMAKE_INSTALL_PREFIX=/usr/local -D OPENCV_EXTRA_MODULES_PATH=/opencv_contrib-3.4.3/modules -D INSTALL_PYTHON_EXAMPLES=ON -D BUILD_EXAMPLES=ON -D BUILD_TIFF=ON -D WITH_VTK=OFF ..

/usr/bin/ld: cannot find -lcblas
/usr/bin/ld: cannot find -latlas
collect2: error: ld returned 1 exit status
Makefile:590: recipe for target '.build_release/lib/libcaffe.so.1.0.0' failed
make: *** [.build_release/lib/libcaffe.so.1.0.0] Error 1
// install
sudo apt-get install libcblas-dev
sudo ln -s libatlas.so.3 libatlas.so

碰到Undefined reference to ‘__cudaPopCallConfiguration’问题

可以在caffe 文件夹下/usr/local/cuda/lib64
ldd build/lib/libcaffe.so | grep cudart
查看cuda 对应so版本
如果不对
需要修改
/usr/lib/x86_64-linux-gnu
中所有 cudaart
替换为
/usr/local/cuda/lib64
中所有cudaart

sudo cp libcudart.so.10.0 /usr/lib/x86_64-linux-gnu/
ting@cidi:/usr/local/cuda/lib64$ sudo cp libcudart.so.10.0.130 /usr/lib/x86_64-linux-gnu/

Caffe添加自定义层，如yolov3使用的层

4: ROS下python或者yaml参数

报错：

Traceback (most recent call last):
  File "/opt/ros/kinetic/bin/roscore", line 69, in <module>
    import roslaunch
  File "/opt/ros/kinetic/lib/python2.7/dist-packages/roslaunch/__init__.py", line 51, in <module>
    from . import param_dump as roslaunch_param_dump
  File "/opt/ros/kinetic/lib/python2.7/dist-packages/roslaunch/param_dump.py", line 40, in <module>
    import roslaunch.config
  File "/opt/ros/kinetic/lib/python2.7/dist-packages/roslaunch/config.py", line 45, in <module>
    import rospkg.distro
  File "/usr/lib/python2.7/dist-packages/rospkg/distro.py", line 44, in <module>
    import yaml
  File "/usr/local/lib/python3.5/dist-packages/yaml/__init__.py", line 284
    class YAMLObject(metaclass=YAMLObjectMetaclass):
                              ^
SyntaxError: invalid syntax

你一定要搞清楚的是PYTHONPATH这个环境变量。遇到的很多导入包的错误都是因为PYTHONPATH没有设置好。例如，要运行roscore会需要导入yaml这个python包，如果你PYTHONPATH没有设置Python2的路径，就会导致roscore引入Python3的yaml的包，造成错误，如下。总而言之，一定要搞清楚PYTHONPATH怎么设置，需要用到哪些路径，不需要用到哪些路径，哪些路径放在前面，哪些放在后面。PS:因为貌似只要一搜索到名字相同的包，系统就会直接引入，不管是版本2还是3；所以如果一个软件包既有2也有3的版本，那么一定要考虑好路径的排列顺序。

解决办法：

export PYTHONPATH="/opt/ros/kinetic/lib/python2.7/dist-packages"

5: TensorRT下配置caffe和onnx运行环境

# TensorRT
find_path(TENSORRT_INCLUDE_DIR NvInfer.h
  HINTS ${
    TENSORRT_ROOT}
  PATH_SUFFIXES include)
MESSAGE(STATUS "Found TensorRT headers at ${TENSORRT_INCLUDE_DIR}")
find_library(TENSORRT_LIBRARY_INFER nvinfer
    HINTS ${
    TENSORRT_ROOT}
  PATH_SUFFIXES lib lib64 lib/x64)
find_library(TENSORRT_LIBRARY_INFER_PLUGIN nvinfer_plugin
    HINTS  ${
    TENSORRT_ROOT}
  PATH_SUFFIXES lib lib64 lib/x64)
find_library(TENSORRT_LIBRARY_PARSER nvparsers 
    HINTS  ${
    TENSORRT_ROOT}
  PATH_SUFFIXES lib lib64 lib/x64)
#find_library(TENSORRT_LIBRARY_ONNXPARSER nvonnxparser 
#    HINTS  ${TENSORRT_ROOT}
#  PATH_SUFFIXES lib lib64 lib/x64)
set(TENSORRT_LIBRARY ${
    TENSORRT_LIBRARY_INFER} ${
    TENSORRT_LIBRARY_INFER_PLUGIN} ${
    TENSORRT_LIBRARY_ONNXPARSER} ${
    TENSORRT_LIBRARY_PARSER})
MESSAGE(STATUS "Find TensorRT libs at ${TENSORRT_LIBRARY}")
find_package_handle_standard_args(
  TENSORRT DEFAULT_MSG TENSORRT_INCLUDE_DIR TENSORRT_LIBRARY)
if(NOT TENSORRT_FOUND)
  message(ERROR
    "Cannot find TensorRT library.")
endif()

需要添加这些

set(TENSORRT_LIBRARY ${
    TENSORRT_LIBRARY_INFER} ${
    TENSORRT_LIBRARY_INFER_PLUGIN} ${
    TENSORRT_LIBRARY_ONNXPARSER} ${
    TENSORRT_LIBRARY_PARSER})

//出现
undefined reference to `nvcaffeparser1::createCaffeParser()
//可能缺少
 ${
    TENSORRT_LIBRARY_PARSER})