Rice Lodging Identification Using Machine Learning

Lim, Win Son (2026) Rice Lodging Identification Using Machine Learning. Final Year Project (Bachelor), Tunku Abdul Rahman University of Management and Technology.

Text Lim Win Son_Thesis.pdf Restricted to Registered users only Download (1MB)

Abstract

Rice lodging significantly impacts crop yield and quality, yet traditional manual assessment is labor-intensive and prone to disputes. This research proposes an automated 2-step identification and calculation system using machine learning on UAV-acquired RGB images. The methodology utilizes the SegFormer-B0 model to first isolate rice paddy fields from background elements like roads and buildings, followed by a secondary model to segment normal and lodged rice. Results demonstrate high precision, with the first model achieving a mean accuracy of 0.99 and a mean Intersection over Union (mIoU) of 0.98. The second model successfully classifies lodged areas with a mean accuracy of 0.88 and a mIoU of 0.78. By providing a standardized, data-driven framework for quantifying damaged areas, this system offers a reliable tool for precision agriculture and objective disaster compensation processes, effectively reducing the time and human error associated with traditional surveys