Simplifying the integration of weed detection into your platform.
The OpenWeedLocator (OWL) is an open-source weed detection device for low-cost,
DIY site-specific weed management. The full project can be found here.
Here, we're making the software that supports the OWL more accesible with an easy-to-use API,
so you can integrate weed detection into your own devices and services more easily
by simply calling GreenonBrown.find().
Currently, owl-tools supports green-on-brown detection on all platforms, and green-on-green
on desktops through the Ultralytics package. Train your only YOLO algorithms and simply integrate
them by providing a path to the trained model.pt.
In the future, we will support plant counting, GPS integration and green-on-green on edge devices (Jetson Nano, Raspberry Pi). If you're nterested in specific features, feel free to raise an issue or pull request.
To install the latest version of openweedlocator-tools simply run:
pip install openweedlocator-tools[desktop]
pip install openweedlocator-tools[rpi]
from owl.viz import webcam, images_and_video
# Run using your webcam
webcam(algorithm='gog', model_path='models/yolov8n.pt') # press escape to exit, add your own model path as required or clone this repository
# or
images_and_video(media_path='path/to/your/media_files')Try it for yourself now with the official Streamlit app!
These examples shows the basic functionality of the detection classes, GreenOnBrown and GreenOnGreen to operate on a video.
from owl.detection import GreenOnBrown
import cv2
VIDEO_PATH = r'media/test_video_1.avi'
ALGORITHM = 'exhsv'
weed_detector = GreenOnBrown()
video_feed = cv2.VideoCapture(VIDEO_PATH)
while True:
ret, frame = video_feed.read()
if not ret:
break
# the weed_detector.find() method returns a tuple with four items
results = weed_detector.find(frame.copy(),
exgMin=30,
exgMax=250,
hueMin=30,
hueMax=90,
brightnessMin=5,
brightnessMax=200,
saturationMin=30,
saturationMax=255,
minArea=1,
show_display=False,
algorithm=ALGORITHM,
invert_hue=False)
contours, bounding_boxes, weed_centres, display_image = results
cv2.imshow('OWL-tput', display_image)
if cv2.waitKey(1) & 0xFF == ord('q'):
breakInstead of instantiating your own GreenOnBrown() instance, you can make the most of
owl.utils.io.get_weed_detector() and owl.utils.io.setup_and_run_detector() functions. The setup_and_run_detector() function
makes it easy to use the inbuilt owl.utils.config.CONFIG_DAY_SENSITIVITY_1, to avoid providing all
variables yourself. For example:
from owl.utils.io import get_weed_detector, load_config, setup_and_run_detector
import cv2
CONFIG_NAME = "CONFIG_DAY_SENSITIVITY_1"
ALGORITHM = 'exhsv'
MODEL_PATH = None
VIDEO_PATH = r'media/test_video_1.avi'
PLATFORM = 'desktop'
weed_detector = get_weed_detector(algorithm=ALGORITHM, model_path=MODEL_PATH, platform='desktop')
config = load_config(CONFIG_NAME)
config.update({"algorithm": f"{ALGORITHM}"})
video_feed = cv2.VideoCapture(VIDEO_PATH)
while True:
ret, frame = video_feed.read()
if not ret:
break
_, _, _, display_image = setup_and_run_detector(weed_detector=weed_detector,
frame=frame.copy(),
config=config)
cv2.imshow('OWL-tput', display_image)
if cv2.waitKey(1) & 0xFF == ord('q'):
break
cv2.destroyAllWindows()If you use this software in your research, please consider citing the original work:
@article{Coleman2022,
author = {Coleman, Guy and Salter, William and Walsh, Michael},
doi = {10.1038/s41598-021-03858-9},
issn = {2045-2322},
journal = {Scientific Reports},
number = {1},
pages = {170},
title = {{OpenWeedLocator (OWL): an open-source, low-cost device for fallow weed detection}},
url = {https://doi.org/10.1038/s41598-021-03858-9},
volume = {12},
year = {2022}
}