wpddmcmc/detection_yolo

detection_yolo

---

Mingcong Chen 03/04/2019

Final report related code

---

1. Introduction

object detecting

2.Environment

System Dependency: Ubuntu16.04

Software Dependency: Cmake2.8 or upper, C++11, OpenCV3.4.0(only), darknet

Hardware Environment:

Computer1: CPU:i7-4720HQ GPU:Nvidia GTX960M-2GB RAM:12GB
Computer2: CPU:i7-8750H GPU:Nvidia GTX1060-6GB RAM:8GB

3.File Structure

src: source code file **include: **head file **build: ** compile file **build/data: ** yolo data list file **yolov3-tiny: **yolo network file **video: **test video

4.Installation

i. Make sure your Software Dependency has been installed

ii.Configure compile

Configure CMakeLists.txt .

gedit CMakeLists.txt

if you have OpenCV version more than one, uncommit and change line8,set the build path which one you want to use.

#if u have OpenCV version more than one, set the build path which one u want to use
set(OpenCV_DIR "YOUR_PATH")

Ex:

#if u have OpenCV version more than one, set the build path which one u want to use
set(OpenCV_DIR "/home/test/app/opencv-3.4.0/build/")

In line50,51, configure the darknet path and cuda path in include_directories

include_directories (
    ${OpenCV_INCLUDE_DIRS}
    /usr/local/include
    /usr/include 
    ${CMAKE_CURRENT_SOURCE_DIR}/include
#darknet path
    YOUR_PATH/include
    YOUR_PATH/src
#cuda path
    YOUR_PATH/include    
)

Ex:

include_directories (
    ${OpenCV_INCLUDE_DIRS}
    /usr/local/include
    /usr/include 
    ${CMAKE_CURRENT_SOURCE_DIR}/include
#darknet path
    /home/test/app/darknet/include
    /home/test/app/darknet/src
#cuda path
    /usr/local/cuda-9.0/include    
)

In line62, configure the darknet library path in target_link_libraries

target_link_libraries(autocar
    ${OpenCV_LIBS}
    /usr/lib
    /usr/local/lib
    ${DEPENDENCIES}
    #darknet lib path
    YOUR_PATH/libdarknet.so
)

Ex:

target_link_libraries(autocar
    ${OpenCV_LIBS}
    /usr/lib
    /usr/local/lib
    ${DEPENDENCIES}
    #darknet lib path
    /home/test/app/darknet/libdarknet.so
)

iii.Compile

Entry build/ and compile

cd build
cmake ..
make

iv.Run

./autocar

5.Using Note

i.Code change

if you changed the code , then need to maketo compile again.

ii.Code file change

If any file changes(add, delete, rename), youneed to cmake .. and makein build/

iii. Camera using

In source file ImageConsProd.cpp line31, uncommit #define USE_CAMERA. Then you can use camera (default 0, you can chage camera index at line49 VideoCapture cap(0);), after that you need to make to compile

iv.Video file using

In source file ImageConsProd.cpp line31, commit #define USE_CAMERA. Then you can use video file, after that you need to make to compile

The choosing of video file. Put video file into video/ folder, then change the video name you want to use in line4 of param_config.xml in param/. Note: just save change, no need to re-compile

v.Debug Mode

In Debug Mode, detect keyboard after load video or camera image. Each frame load with each click of keyboard.

Debug Mode ON/OFF

Change line3 of param_config.xml in param/, 1 ON, 0 OFF.

<?xml version="1.0"?>
<opencv_storage>
<debug_mode>1</debug_mode>
<video_name>vedioname.mp4</video_name>
</opencv_storage>