This package provides a ROS wrapper for YOLOv5 based on PyTorch-YOLOv5.
Authors: Vasileios Vasilopoulos ([email protected]), Georgios Pavlakos ([email protected]) Adapted by: Raghava Uppuluri
- Have conda installed
- Have ROS installed + setup
- Download the prerequisites for this package, navigate to the package folder and run:
# Ensure your conda environment is activated
conda install -f requirements.txt
Clone the repo into your src
folder of your catkin_ws
.
git clone https://github.com/raghavauppuluri13/yolov5_pytorch_ros.git
Navigate to your catkin workspace and run:
catkin build yolov5_pytorch_ros
# adds package to your path
source ~/catkin_ws/devel/setup.bash
To maximize portability, create a separate package and launch file. Add your weights into a weights
folder of that package.
catkin_create_pkg my_detector
mkdir weights
mkdir launch
# Add weights
# Don't forget to build and source after
Then, add the following to mydetector.launch
in the launch folder:
<launch>
<include file="$(find yolov5_pytorch_ros)/launch/detector.launch">
<!-- Camera topic and weights, config and classes files -->
<arg name="image_topic" value="/camera/image_raw"/>
<!-- Absolute path to weights file (change this) -->
<arg name="weights_name" value="$(find my_detector)/weights/weights.pt"/>
<!-- Published topics -->
<arg name="publish_image" value="true"/>
<arg name="detected_objects_topic" value="detected_objects_in_image"/>
<arg name="detections_image_topic" value="detections_image_topic"/>
<!-- Detection confidence -->
<arg name="confidence" value="0.7"/>
</include>
</launch>
Finally, run the detector:
roslaunch my_detector mydetector.launch
Should get something like this when viewed from rviz
-
image_topic
(string)Subscribed camera topic.
-
weights_name
(string)Weights to be used from the models folder.
-
publish_image
(bool)Set to true to get the camera image along with the detected bounding boxes, or false otherwise.
-
detected_objects_topic
(string)Published topic with the detected bounding boxes.
-
detections_image_topic
(string)Published topic with the detected bounding boxes on top of the image.
-
confidence
(float)Confidence threshold for detected objects.
-
image_topic
(sensor_msgs::Image)Subscribed camera topic.
-
detected_objects_topic
(yolov3_pytorch_ros::BoundingBoxes)Published topic with the detected bounding boxes.
-
detections_image_topic
(sensor_msgs::Image)Published topic with the detected bounding boxes on top of the image (only published if
publish_image
is set to true).
The YOLO methods used in this software are described in the paper: You Only Look Once: Unified, Real-Time Object Detection.
If you are using this package, please add the following citation to your publication:
@misc{vasilopoulos_pavlakos_yolov3ros_2019,
author = {Vasileios Vasilopoulos and Georgios Pavlakos},
title = {{yolov3_pytorch_ros}: Object Detection for {ROS} using {PyTorch}},
howpublished = {\url{https://github.com/vvasilo/yolov3_pytorch_ros}},
year = {2019},
}