YOLO V3学习（3）——基于darknet ros的目标检测 基于gpu版本

本机已经安装GPU显卡，并下载了Cuda、Cudnn。GPU配置为1660Ti

1.darknet编译
下载darknet：
git clone https://github.com/AlexeyAB/darknet.git
修改相应的makefile文件
主要的修改部分：

GPU=1
CUDNN=1
CUDNN_HALF=0
OPENCV=1
AVX=0
OPENMP=0
LIBSO=0
ZED_CAMERA=0
ZED_CAMERA_v2_8=0

# set GPU=1 and CUDNN=1 to speedup on GPU
# set CUDNN_HALF=1 to further speedup 3 x times (Mixed-precision on Tensor Cores) GPU: Volta, Xavier, Turing, Ampere, Ada and higher
# set AVX=1 and OPENMP=1 to speedup on CPU (if error occurs then set AVX=0)
# set ZED_CAMERA=1 to enable ZED SDK 3.0 and above
# set ZED_CAMERA_v2_8=1 to enable ZED SDK 2.X

USE_CPP=0
DEBUG=0

#ARCH= -gencode arch=compute_50,code=[sm_50,compute_50] \
#      -gencode arch=compute_52,code=[sm_52,compute_52] \
#	    -gencode arch=compute_61,code=[sm_61,compute_61]

OS := $(shell uname)

# Naming confusion with recent RTX cards.
# "NVIDIA Quadro RTX x000" and T1000/Tx00 are Turing Architecture Family with Compute Capability of 7.5
# "NVIDIA RTX Ax000" are Ampere Architecture Family with Compute Capability of 8.6
# NVIDIA "RTX x000 Ada" are Ada Lovelace Architecture Family with Compute Capability of 8.9
# Source https://developer.nvidia.com/cuda-gpus

# KEPLER, GeForce GTX 770, GTX 760, GT 740
# ARCH= -gencode arch=compute_30,code=sm_30

# MAXWELL, GeForce GTX 950, 960, 970, 980, 980 Ti, "GTX" Titan X
# ARCH= -gencode arch=compute_52,code=sm_52

# Jetson TX1, Tegra X1, DRIVE CX, DRIVE PX, Jetson Nano (2GB, 4GB)
# ARCH= -gencode arch=compute_53,code=[sm_53,compute_53]

# GP100/Tesla P100 - DGX-1
# ARCH= -gencode arch=compute_60,code=sm_60

# PASCAL, GTX 10x0, GTX 10x0 Ti, Titan Xp, Tesla P40, Tesla P4
ARCH= -gencode arch=compute_61,code=[sm_61,compute_61]

# For Jetson TX2, Jetson Nano TX2 or Drive-PX2 uncomment:
# ARCH= -gencode arch=compute_62,code=[sm_62,compute_62]

# Tesla V100
# ARCH= -gencode arch=compute_70,code=[sm_70,compute_70]

# Jetson XAVIER, XAVIER NX
# ARCH= -gencode arch=compute_72,code=[sm_72,compute_72]

# GeForce Titan RTX, RTX 20x0, RTX 20x0 Ti, Quadro RTX x000, Tesla T4, XNOR Tensor Cores
# ARCH= -gencode arch=compute_75,code=[sm_75,compute_75]

# Tesla A100 (GA100), DGX-A100, A30, A100, RTX 3080
# ARCH= -gencode arch=compute_80,code=[sm_80,compute_80]

# GeForce RTX 30x0, 30x0 Ti, Tesla GA10x, RTX Axxxx, A2, A10, A16, A40
# ARCH= -gencode arch=compute_86,code=[sm_86,compute_86]

# NOT TESTED, THEORETICAL
# Jetson ORIN, ORIN NX, ORIN NANO
# ARCH= -gencode arch=compute_87,code=[sm_87,compute_87]

# NOT TESTED, THEORETICAL
# GeForce RTX 4070 Ti, 4080, 4090, L4, L40
# ARCH= -gencode arch=compute_89,code=[sm_89,compute_89]

# NOT TESTED, THEORETICAL
# Nvidia H100
# ARCH= -gencode arch=compute_90,code=[sm_90,compute_90]

VPATH=./src/
EXEC=darknet
OBJDIR=./obj/

cd darknet
make
测试：

./darknet detect cfg/yolov3.cfg yolov3.weights data/dog.jpg

2.darknet_ros 编译
下载darknet_ros包和darknet包
darknet_ros:包
cd src
git clone https://github.com/leggedrobotics/darknet_ros
darknet包，这里darknet_ros包里有三个文件夹，直接替换其中的darknet或者删除这个文件夹：
git clone https://github.com/AlexeyAB/darknet
问题一：darknet_ros 会开始下载yolov2-tiny.weights、yolov3.weights和yolov2.weights，但国内下载太慢，可以直接去CMakelists.txt里注释掉。
解决如下：

# Download yolov2-tiny.weights
set(PATH "${CMAKE_CURRENT_SOURCE_DIR}/yolo_network_config/weights")
set(FILE "${PATH}/yolov2-tiny.weights")
message(STATUS "Checking and downloading yolov2-tiny.weights if needed ...")
if (NOT EXISTS "${FILE}")
  message(STATUS "... file does not exist. Downloading now ...")
#  execute_process(COMMAND wget -q https://github.com/leggedrobotics/darknet_ros/releases/download/1.1.4/yolov2-tiny.weights -P ${PATH})
endif()

# Download yolov3.weights
set(FILE "${PATH}/yolov3.weights")
message(STATUS "Checking and downloading yolov3.weights if needed ...")
if (NOT EXISTS "${FILE}")
  message(STATUS "... file does not exist. Downloading now ...")
#  execute_process(COMMAND wget -q https://github.com/leggedrobotics/darknet_ros/releases/download/1.1.4/yolov3.weights -P ${PATH})
endif()

#############
## Testing ##
#############

if(CATKIN_ENABLE_TESTING)
  # Download yolov2.weights
  set(PATH "${CMAKE_CURRENT_SOURCE_DIR}/yolo_network_config/weights")
  set(FILE "${PATH}/yolov2.weights")
  message(STATUS "Checking and downloading yolov2.weights if needed ...")
  if (NOT EXISTS "${FILE}")
    message(STATUS "... file does not exist. Downloading now ...")
#    execute_process(COMMAND wget -q https://github.com/leggedrobotics/darknet_ros/releases/download/1.1.4/yolov2.weights -P ${PATH})
  endif()

  find_package(rostest REQUIRED)

问题二：需要更改相应的gencode,具体可以参考darknet中各种显卡对应。我这里是1660Ti.

#ARCH= -gencode arch=compute_50,code=[sm_50,compute_50] \
#      -gencode arch=compute_52,code=[sm_52,compute_52] \
#	    -gencode arch=compute_61,code=[sm_61,compute_61]

OS := $(shell uname)

# Naming confusion with recent RTX cards.
# "NVIDIA Quadro RTX x000" and T1000/Tx00 are Turing Architecture Family with Compute Capability of 7.5
# "NVIDIA RTX Ax000" are Ampere Architecture Family with Compute Capability of 8.6
# NVIDIA "RTX x000 Ada" are Ada Lovelace Architecture Family with Compute Capability of 8.9
# Source https://developer.nvidia.com/cuda-gpus

# KEPLER, GeForce GTX 770, GTX 760, GT 740
# ARCH= -gencode arch=compute_30,code=sm_30

# MAXWELL, GeForce GTX 950, 960, 970, 980, 980 Ti, "GTX" Titan X
# ARCH= -gencode arch=compute_52,code=sm_52

# Jetson TX1, Tegra X1, DRIVE CX, DRIVE PX, Jetson Nano (2GB, 4GB)
# ARCH= -gencode arch=compute_53,code=[sm_53,compute_53]

# GP100/Tesla P100 - DGX-1
# ARCH= -gencode arch=compute_60,code=sm_60

# PASCAL, GTX 10x0, GTX 10x0 Ti, Titan Xp, Tesla P40, Tesla P4
ARCH= -gencode arch=compute_61,code=[sm_61,compute_61]

# For Jetson TX2, Jetson Nano TX2 or Drive-PX2 uncomment:
# ARCH= -gencode arch=compute_62,code=[sm_62,compute_62]

# Tesla V100
# ARCH= -gencode arch=compute_70,code=[sm_70,compute_70]

# Jetson XAVIER, XAVIER NX
# ARCH= -gencode arch=compute_72,code=[sm_72,compute_72]

# GeForce Titan RTX, RTX 20x0, RTX 20x0 Ti, Quadro RTX x000, Tesla T4, XNOR Tensor Cores
# ARCH= -gencode arch=compute_75,code=[sm_75,compute_75]

# Tesla A100 (GA100), DGX-A100, A30, A100, RTX 3080
# ARCH= -gencode arch=compute_80,code=[sm_80,compute_80]

# GeForce RTX 30x0, 30x0 Ti, Tesla GA10x, RTX Axxxx, A2, A10, A16, A40
# ARCH= -gencode arch=compute_86,code=[sm_86,compute_86]

# NOT TESTED, THEORETICAL
# Jetson ORIN, ORIN NX, ORIN NANO
# ARCH= -gencode arch=compute_87,code=[sm_87,compute_87]

# NOT TESTED, THEORETICAL
# GeForce RTX 4070 Ti, 4080, 4090, L4, L40
# ARCH= -gencode arch=compute_89,code=[sm_89,compute_89]

# NOT TESTED, THEORETICAL
# Nvidia H100
# ARCH= -gencode arch=compute_90,code=[sm_90,compute_90]

# Find CUDA
find_package(CUDA QUIET)
if (CUDA_FOUND)
  find_package(CUDA REQUIRED)
  message(STATUS "CUDA Version: ${CUDA_VERSION_STRINGS}")
  message(STATUS "CUDA Libararies: ${CUDA_LIBRARIES}")
  set(
    CUDA_NVCC_FLAGS
    ${CUDA_NVCC_FLAGS};
    -O3
    -gencode arch=compute_61,code=[sm_61,compute_61]
  )
  add_definitions(-DGPU)
else()
  list(APPEND LIBRARIES "m")
endif()

这里使用kitti数据集：

<?xml version="1.0" encoding="utf-8"?>

<launch>
  <!-- Console launch prefix -->
  <arg name="launch_prefix" default=""/>
  <arg name="image" default="/kitti/camera_color_left/image_raw" />

  <!-- Config and weights folder. -->
  <arg name="yolo_weights_path"          default="$(find darknet_ros)/yolo_network_config/weights"/>
  <arg name="yolo_config_path"           default="$(find darknet_ros)/yolo_network_config/cfg"/>

  <!-- ROS and network parameter files -->
  <arg name="ros_param_file"             default="$(find darknet_ros)/config/ros.yaml"/>
  <arg name="network_param_file"         default="$(find darknet_ros)/config/yolov3.yaml"/>

  <!-- Load parameters -->
  <rosparam command="load" ns="darknet_ros" file="$(arg ros_param_file)"/>
  <rosparam command="load" ns="darknet_ros" file="$(arg network_param_file)"/>

  <!-- Start darknet and ros wrapper -->
  <node pkg="darknet_ros" type="darknet_ros" name="darknet_ros" output="screen" launch-prefix="$(arg launch_prefix)">
    <param name="weights_path"          value="$(arg yolo_weights_path)" />
    <param name="config_path"           value="$(arg yolo_config_path)" />
    <remap from="camera/rgb/image_raw"  to="$(arg image)" />
  </node>

 <!--<node name="republish" type="republish" pkg="image_transport" output="screen" 	args="compressed in:=/front_camera/image_raw raw out:=/camera/image_raw" /> -->
</launch>

另外需要说明一点的是从如下链接下载文件放入/darknet_ros/darknet_ros/yolo_network_config/weights文件中

https://pjreddie.com/media/files/yolov2.weights
https://pjreddie.com/media/files/yolov2-tiny.weights
https://pjreddie.com/media/files/yolov3.weights
https://pjreddie.com/media/files/yolov3-tiny.weights

下一节会学习数据集的标注。