Enhancing grain moisture prediction in multiple crop seasons using domain adaptation AI

Ming-Der Yang, Yu-Chun Hsu, Tsai-Ting Liu, Han-Hui Huang

Abstract: Highlights•Developing a domain adaptation (DA) workflow for predicting rice grain moisture content (GMC) in multiple crop seasons.•Designing a real-time light-weight detector using U2-Net and YOLOv7 with a prediction error less than 1 % GMC.•Achieving a 54.66 % improvement in DA-modified GMC prediction with only 0.1 % target domain data.•Generating a rice maturity map through supervised and unsupervised DA with prediction errors within 1.2 % GMC.