Yolov5 tensorrt acceleration

Video tutorial

https://www.bilibili.com/video/BV113411J7nk?p=1

Github Warehouse address

https://github.com/Monday-Leo/Yolov5_Tensorrt_Win10

Environmental Science

Tensorrt 8.2.1.8
Cuda 10.2 Cudnn 8.2.1
Cuda Install Links :http://t.csdn.cn/AXNks

Opencv 3.4.6
Cmake 3.17.1
Cmake Installation tutorial :http://t.csdn.cn/f2TYW

VS 2017
GTX1650

Opencv Configuration method
1、 stay OpenCV The official website download is applicable to Windows Platform 3.4.6 edition

2、 Run the downloaded executable , take OpenCV Extract to the specified directory , for example D:\propencv

3、 My computer -> attribute -> Advanced system setup -> environment variable , Find... In the system variable Path（ If not , Create your own ）, And double-click edit , take opencv Fill in the path and save , Such as D:\opencv\build\x64\vc15\bin

Tensorrt Configuration method
1、 stay tensorrt The official website download is applicable to Windows Version of the platform

2、 take TensorRT/lib All of them lib Copied to the cuda/v10.2/lib/x64 Next , take TensorRT/lib All of them dll Copied to the cuda/v10.2/bin Next , take TensorRT/include All of them .h File copy to cuda/v10.2/include Next

3、 My computer -> attribute -> Advanced system setup -> environment variable , Find... In the system variable Path（ If not , Create your own ）, And double-click edit , take TensorRT/lib Fill in the path and save , Such as G:\TensorRT-8.2.1.8\lib

Open the warehouse CMakeLists.txt, modify Opencv、Tensorrt、dirent.h The catalog of , among dirent.h In this warehouse include in , Must be an absolute path . modify arch=compute_75;code=sm_75, Reference resources https://developer.nvidia.com/zh-cn/cuda-gpus,GPU by GTX1650, Ability to calculate 7.5, So this is set to arch=compute_75;code=sm_75.

cmake_minimum_required(VERSION 2.6)

project(yolov5) 

#change to your own path
##################################################
set(OpenCV_DIR "D:\\opencv\\build")
set(TRT_DIR "D:\\TensorRT-8.2.1.8")
set(Dirent_INCLUDE_DIRS "D:\\Pycharm.project\\Yolov5_Tensorrt_Win10-master\\include")
##################################################

add_definitions(-std=c++11)
add_definitions(-DAPI_EXPORTS)
option(CUDA_USE_STATIC_CUDA_RUNTIME OFF)
set(CMAKE_CXX_STANDARD 11)
set(CMAKE_BUILD_TYPE Debug)

set(THREADS_PREFER_PTHREAD_FLAG ON)
find_package(Threads)

# setup CUDA
find_package(CUDA REQUIRED)
message(STATUS "    libraries: ${CUDA_LIBRARIES}")
message(STATUS "    include path: ${CUDA_INCLUDE_DIRS}")
include_directories(${CUDA_INCLUDE_DIRS})
include_directories(${Dirent_INCLUDE_DIRS}) 

#change to your GPU own compute_XX
###########################################################################################
set(CUDA_NVCC_FLAGS ${CUDA_NVCC_FLAGS};-std=c++11;-g;-G;-gencode;arch=compute_75;code=sm_75)
###########################################################################################

####
enable_language(CUDA)  # add this line, then no need to setup cuda path in vs
####
include_directories(${PROJECT_SOURCE_DIR}/include)
include_directories(${TRT_DIR}\\include)

# -D_MWAITXINTRIN_H_INCLUDED for solving error: identifier "__builtin_ia32_mwaitx" is undefined
set(CMAKE_CXX_FLAGS "${CMAKE_CXX_FLAGS} -std=c++11 -Wall -Ofast -D_MWAITXINTRIN_H_INCLUDED")

# setup opencv
find_package(OpenCV QUIET
    NO_MODULE
    NO_DEFAULT_PATH
    NO_CMAKE_PATH
    NO_CMAKE_ENVIRONMENT_PATH
    NO_SYSTEM_ENVIRONMENT_PATH
    NO_CMAKE_PACKAGE_REGISTRY
    NO_CMAKE_BUILDS_PATH
    NO_CMAKE_SYSTEM_PATH
    NO_CMAKE_SYSTEM_PACKAGE_REGISTRY
)

message(STATUS "OpenCV library status:")
message(STATUS "    version: ${OpenCV_VERSION}")
message(STATUS "    libraries: ${OpenCV_LIBS}")
message(STATUS "    include path: ${OpenCV_INCLUDE_DIRS}")

include_directories(${OpenCV_INCLUDE_DIRS})
link_directories(${TRT_DIR}\\lib)

add_executable(yolov5 ${PROJECT_SOURCE_DIR}/yolov5.cpp ${PROJECT_SOURCE_DIR}/yololayer.cu ${PROJECT_SOURCE_DIR}/yololayer.h ${PROJECT_SOURCE_DIR}/preprocess.cu) 

target_link_libraries(yolov5 "nvinfer" "nvinfer_plugin")  
target_link_libraries(yolov5 ${OpenCV_LIBS})     
target_link_libraries(yolov5 ${CUDA_LIBRARIES})  
target_link_libraries(yolov5 Threads::Threads)     

download yolov5 6.0 Source code and this warehouse code

git clone -b v6.0 https://github.com/ultralytics/yolov5.git
git clone https://github.com/Monday-Leo/Yolov5_Tensorrt_Win10

Generate WTS Model

Put... In the warehouse gen_wts.py And just downloaded yolov5s.pt Copy to yolov5 6.0 Under the directory of

function gen_wts.py

python gen_wts.py -w yolov5s.pt -o yolov5s.wts

Generate wts file

Cmake The process

Create a new one in the directory build Folder

open Cmake, Select this warehouse directory , And the new build Catalog , Then click on the bottom left configure Button .

Choose your own Visual Studio edition , Such as 2017, In the second box, select x64, After clicking on finish

Running results

*

In turn, click Generate and Open Project

compile

Place the top of the interface Debug Change it to Release, Right click yolov5 project , Click regenerate .

Compilation successfully opened build/Release, You can see the generated exe Executable file .

What will be generated in the first step yolov5s.wts Copy the model to exe In the folder of , Open in this directory cmd Input

yolov5 -s yolov5s.wts yolov5s.engine s

The normal operation , At this time, the program will wts Convert to engine Serialization model , Need to wait for the forecast 10-20 About minutes . Generate engine After completion , Will appear under the folder yolov5s.engine Model . Put the pictures Copy folder to exe Under the folder , Try to predict correctly , Input ：

yolov5 -d yolov5s.engine ./pictures