Xiangji Meng

Conducted research under the Hokkaido Data-Driven Convergent Research and Digital Human Resource Development Program (D-RED), aiming to advance interdisciplinary research through data science and AI. The project fosters collaboration between academia, industry, and government while leveraging regional strengths to accelerate societal implementation. By promoting data-driven convergent research, the initiative seeks to create new value originating from Hokkaido.

• Developed a deep learning-based multidimensional detection model for plant stomata, integrating features such as aperture area, complex area, guard cell area, density, spatial distribution, stomatal state (open/closed), and maturity. The model achieves high-throughput, high-accuracy, and cost-effective performance.
• To enhance usability, an auto-labeling function was implemented, enabling independent researchers to achieve high model accuracy with minimal training effort.
• Additionally, a machine learning model was trained using multidimensional stomatal data collected by the deep learning system to predict plant growth, future yield, and potential disease risks.