far-planner-with-yolo

FAR Planner version with yolo to detect object

Tested in ubuntu 20.04, ubuntu 18.04, Manjaro (you need to use AUR to build ROS1 noetic(recommend) or melodic)， Deepin20.4 with noetic

Note that the opencv version should higher than 3.3 , 4.5.4 is recommended

opencv的版本要求要大于3.3,最好用4.5.4

you need to build newer opencv from source

我们需要去opencv的官网进行源码编译，链接如下

download source from https://opencv.org/

cd opencv-4.5.4
mkdir build & cd build
cd opencv-4.5.4
cmake -DCMAKE_BUILD_TYPE=RELEASE -DCMAKE_INSTALL_PREFIX=/usr/local -DWITH_TBB=ON -DBUILD_NEW_PYTHON_SUPPORT=ON -DWITH_V4L=ON -DINSTALL_C_EXAMPLES=ON -DINSTALL_PYTHON_EXAMPLES=ON -DBUILD_EXAMPLES=ON -DWITH_QT=ON -DWITH_GTK=ON -DWITH_OPENGL=ON ..
make
sudo make install

yolo cfg, weight and coconame should be downloaded and put at /src/yolo/data, and the default is yolov4-tiny

yolo相关的cfg、weight和coconame要放在/src/yolo/data这个目录下，目前默认的为yolov4-tiny

Setting yolo config file in the yolo/src/detection.cpp, change the absolute path for the cfg, weight and coconame

在yolo/src/detection.cpp里设置cfg、weight、coconame的绝对路径

and then, plesae use catkin_make 执行catkin_make

cd /far-planner-with-yolo
catkin_make
sudo chmod +x run.sh
./run.sh

cfg weight coconames can be download from https://github.com/AlexeyAB/darknet
这个是对应的cfg、weight和coconames的下载连接，请选择合适的进行使用

The original far_planner is here, please check the url below
https://github.com/MichaelFYang/far_planner