Weedy Rice DSS
An applied research platform for agricultural segmentation and decision support.
Weedy Rice DSS is a cross-season domain-gap study in UAV weedy-rice segmentation, built around a thesis now in final review. It measures the gap between wet- and dry-season imagery, then tests whether unsupervised domain-adaptation methods (FDA, DAFormer and HRDA) close it, under pre-specified decision rules. A dormant product layer (Python backend, React/Vite UI, annotation tooling) was built earlier and is retained but is not part of the research path.
Overview
The Weedy Rice Decision Support System (DSS) is a cross-season domain-gap study in UAV weedy-rice segmentation, built around a thesis now in final review. It began as an applied platform for AI-driven agricultural image segmentation; that earlier product layer is retained but dormant.
Architecture
The research path is a Python training and evaluation stack for semantic segmentation models, using GPU training, pre-specified statistical tests and decision rules written down before the key reruns and analyses. The earlier product layer (a Python backend, a React/Vite frontend and annotation tooling) is archived and not part of the research path. A prototype serverless pipeline on AWS was built along the way; its production stack has since been torn down.
Problem It Solves
Models trained on imagery from one season often behave differently on another. The study measures the gap between wet- and dry-season imagery and tests whether unsupervised domain adaptation closes it, reporting what it finds either way.
How the Work Is Checked
The work is designed by Peem and implemented with AI coding tools, so nothing is accepted because an agent said it. Decision rules are written down before an experiment runs, and a claim counts only once a separate check has confirmed it against a primary source such as the code, the data, a file hash or the compiled document.
Current State
The thesis is in final review. The earlier product layer is dormant, and the AWS prototype's production stack has been torn down.