Decision intelligence
Weed morphology–guided spray decisions
Overview
Detecting a weed answers where to spray; this study addresses how to spray it by adapting delivery decisions to weed size and growth habit.
Watch demonstrations
- Species
- Marestail · lambsquarters
- Imaging
- Top + side stereo views
- Treatments
- Tip · angle · speed · size
- Response
- Canopy + ground deposition
Weed morphology and adaptive dosing
These demonstrations show the complete greenhouse workflow: controlled image acquisition, morphology extraction, and the software environment used to turn plant measurements into size-aware spray decisions.
Greenhouse gantry imaging system
The automated gantry moves across the plant array to support repeatable top- and side-view imaging throughout plant development.
1 / 3Imaging and experiment-control interface
The custom GUI coordinates three camera views, region-of-interest settings, image capture, calibration, and repeatable greenhouse data collection.
2 / 3Morphology extraction with HSV segmentation
Plant pixels are separated from the background to quantify projected area and morphology features used in weed-size and growth-habit characterization.
3 / 3Research Challenge
A binary spray/no-spray decision ignores plant size, architecture, and the amount of spray required for effective coverage.
Engineering Approach
- Capture repeatable multiview plant imagery with a dual-axis gantry.
- Extract projected area, shape descriptors, and growth-habit features.
- Classify plant size and architecture for treatment selection.
- Evaluate size-linked spray volume, nozzle angle, travel speed, and deposition.
Experimental design
- Contrasting weed species
- 2
- Size classes
- 3
- Treatments per plant
- 18
- Plant-level deposition records
- 324
Depth-Calibrated Physical Weed Size Extraction
Significance
The work advances precision spraying from target detection toward biologically informed treatment selection and better weed coverage.