Ramie Yield Estimation Based on UAV RGB Images.

Area of Science:

• Agricultural Science
• Remote Sensing
• Agronomy

Background:

• Traditional ramie yield estimation methods are destructive and labor-intensive.
• Accurate crop monitoring is crucial for effective field management and optimizing agricultural practices.

Discussion:

• Unmanned aerial vehicle (UAV) platforms equipped with RGB cameras can capture detailed crop canopy data throughout the growth cycle.
• Extracted vegetation indices, plant number, and plant height from UAV imagery serve as key phenotypic parameters.
• Structural features like plant number and height are more indicative of ramie yield than spectral features.

Key Insights:

• Plant number emerged as the most influential structural feature for yield monitoring, showing a correlation coefficient of 0.6.
• A multiple linear regression model integrating multiple phenotypic parameters demonstrated superior accuracy (R²=0.66) over stepwise regression.
• The fusion of phenotypic data significantly enhances the precision of ramie yield estimation models.

Outlook:

• This study validates the feasibility of using UAV-based imagery for crop growth monitoring.
• Integrating diverse phenotypic data from UAVs holds promise for improving the accuracy and efficiency of agricultural yield estimations.
• Further research can explore advanced machine learning techniques for even more robust yield prediction models.