Location: Sugarbeet Research
Project Number: 3060-21000-045-029-S
Project Type: Non-Assistance Cooperative Agreement
Start Date: Jun 1, 2024
End Date: May 31, 2025
Objective:
Main objective of this research is to investigate AI-driven segmentation and detection models to improve the accuracy of weed identification.
Approach:
Novel methods will be utilized to investigate th quality of masks generated by Segment Anything Model (SAM) for weed identification. Experiments will be conducted in a automated mask generator to prevent human bias from prompting and AI-driven detection models to extract target weeds with SAM-enhanced training dataset, and the cooperator will carry out the following:
1. Provide expertise in planning and conducting training data experiments and AI model optimization.
2. Ensure the algorithm design and methods are scientifically sound and are completed in a timely manner for the outlined sugarbeet weed control experiments.
3. Ensure the AI training data and models are analyzed and interpreted accurately, and results are reported in timely manner.
4. Contribute preparation of a final report to distribute the results.
