Matthew Kutugata retweeted
The OpenSourceAg newsletter is here!
As a starting point for an open-source (OS) community in ag, it will feature:
🌱summaries of latest tech for data, algos, field use
🌱stories from OS founders
🌱projects
🌱community comments
Signup here:
opensourceag.beehiiv.com/sub…
Matthew Kutugata retweeted
About 50 years ago, the tech behind reflectance-based 'green-on-brown' sprayers was published.
Since then a lot (!) has changed.
We went through ~175 papers from the last 50 years of weed recognition research for this review.
cambridge.org/core/journals/…
Some more weed action from @PSA_Network large image repository currently in the works. Shout out to @savadikarc for helping develop the pipeline. You miss us yet?
Matthew Kutugata retweeted
When they saw record numbers of young people coming out for @BetoORourke they did this:
🚨ALERT: Brazos County, Texas, refuses to reopen an on-campus early voting site at Texas A&M University in College Stations, the largest campus in the state, for the 2022 midterm elections.
Matthew Kutugata retweeted
New student organization focused on cannabis future in Texas kicks off with more than 100 participants.
A little weed from the image repository currently in the works with folks from @PSA_Network and @NCStateCALS
Matthew Kutugata retweeted
RootPainter: Deep Learning Segmentation of Biological Images with Corrective Annotation. Now in @NewPhyt nph.onlinelibrary.wiley.com/…
Matthew Kutugata retweeted
Will be looking for MS/PhD students, interested in biology, ecology, and integrated managements of weeds. Research ideas and approaches will be diverse and multidisciplinary. Current email id aniruddha.maity@ag.tamu.edu
Matthew Kutugata retweeted
Guest contributor Matthew Klein discusses challenges of being a queer teacher, especially within the context of recent anti-LGBTQ+ legislation.
thebatt.com/opinion/opinion-…
Matthew Kutugata retweeted
New: Importing COCO JSON file can import Weed-AI Broadleaf Weeds in Common Couch dataset by @GeezaColeman.
weed-ai.sydney.edu.au/datase…
Matthew Kutugata retweeted
Images are everything in #deeplearning!
E.g. #ImageNet has >14 M images in 20k classes. Nothing of the same scale exists for weeds (@WeedAI_news is trying!) but there are open datasets.
For weed recognition, here's a 🧵of all weeds datasets I've found.
📸🌱🧵
Matthew Kutugata retweeted
This precision camera system was developed by Danish scientists for high-speed mapping and controlling of weeds in forage crops. The PSA and GROW teams are collaborating to test and develop this technology for mapping cover crops in the US. @USDA_NIFA @USDA_ARS @GetRidOfWeeds
Matthew Kutugata retweeted
Spent the last 2 months working with @Abe_404 @DIKU_Institut and his open-source software RootPainter. Aim was to extract cotton/Palmer amaranth from some glasshouse photos collected in Texas.
Where colour segmentation failed, RootPainter helped with corrective annotation.
Matthew Kutugata retweeted
Want to learn how to build an #OpenWeedLocator?
We're holding a live (online) assembly & demo on 4th May, 12pm (AEST)!
Step-by-step walkthrough on hardware assembly & software installation with myself & @williamtsalter.
@SiaSydney @Sydney_Science
eventbrite.com.au/e/openweed…
Matthew Kutugata retweeted
🚨 New @WeedAI_news dataset available!
Precision Sustainable Ag 2021 OpenCV Competition from @mattkutu and co:
🌱728 images
🌱7 weed species
🌱3,444 bounding box labels
Access it here:
weed-ai.sydney.edu.au/datase…
Using images and some tools from @plantcv for collecting morphology data on annual bluegrass, an invasive weed found on every continent including Antarctica. Working with @HeyOsburn and his research at @AggieTurf on this one.

ALT Overhead image of annual bluegrass in a research nursery. Notice the barrel distortion around the edges that looks like the image is rounding. This is caused by the camera lens but can be corrected using some simple calibration techniques.

ALT Binary segmentation mask of our target weed that we want to take data from. Coupled with calibration data, this image alone is enough to tell us about about size, shape, area, perimeter, solidity, and other useful pieces of information. Multiply this by 400 plants, 10 subtypes, and taken across 11 time periods, we have a large dataset ready to be analyzed. And all we did was take pictures.
Matthew Kutugata retweeted
The 1st chapter of my PhD has been awarded as the Weed Science journal's best paper for 2022.
cup.org/3LWgTYs
Thanks to @WorldofWeeds @WSSA_GSO
























