Skip to content

Latest commit

 

History

History
74 lines (56 loc) · 2.1 KB

intro_to_yolo.md

File metadata and controls

74 lines (56 loc) · 2.1 KB
Raw

Introduction to YOLO/Darknet Image Recognition

Video Tutorial at https://youtu.be/SiVexS6Zrr8

Install CUDA

Cuda is a library that allows programs to take advantage of your GPU as a computing resource. YOLO will run without Cuda, but the algorithm is up to 500 x more quick with Cuda. To install Cuda, run

sudo apt install nvidia-cuda-toolkit

Ubuntu 18.04

Clone Darknet/YOLO

Clone the darknet repo into our catkin_ws

cd ~/catkin_ws/src
git clone --recursive https://github.com/leggedrobotics/darknet_ros.git

Build Darknet

catkin build -DCMAKE_BUILD_TYPE=Release

on Ubuntu 18.04 you may need to run

catkin build -DCMAKE_BUILD_TYPE=Release -DCMAKE_C_COMPILER=/usr/bin/gcc-6

Ubuntu 20.04

cd ~/catkin_ws/src
git clone https://github.com/kunaltyagi/darknet_ros.git
cd darknet_ros/
git checkout opencv4
git submodule update --init --recursive

Build Darknet

catkin build -DCMAKE_BUILD_TYPE=Release

if you run into errors try running the following

catkin build -DCMAKE_BUILD_TYPE=Release -DCMAKE_C_COMPILER=/usr/bin/gcc-8

Configure YOLO/Darknet

in the file ros.yaml specifies ros parameters. You can find this file under darknet_ros/darknet_ros/config. You will need to change the image topic from /camera/rgb/image_raw to

/webcam/image_raw

The file darknet_ros.launch will launch the darknet/yolo ros node. You can find this file under darknet_ros/darknet_ros/launch

in this file you can choose which version of yolo you would like to run by changing

<arg name="network_param_file"         default="$(find darknet_ros)/config/yolov2-tiny.yaml"/>

the options are as follows

  • yolov1: Not recommended. this model is old
  • yolov2: more accurate, and faster.
  • yolov3: about as fast as v2, but more accurate. Yolo v3 has a high GPU ram requirement to train and run. If your graphics card does not have enough ram, use yolo v2
  • tiny-yolo: Very fast yolo model. Would recommend for application where speed is most important. Works very well on Nvidia Jetson

References

https://pjreddie.com/darknet/yolo/