SLPA-Net: A Real-Time Recognition Network for Intelligent Stomata Localization and Phenotypic Analysis

Area of Science:

• Plant Science
• Agricultural Engineering
• Computer Vision

Background:

• Manual measurement of plant stomatal traits is time-consuming and labor-intensive.
• Accurate stomatal phenotype data is crucial for improving crop water use efficiency, stress resistance, and yield.

Purpose of the Study:

• To develop a high-throughput method for stomatal localization and phenotypic analysis.
• To introduce SLPA-Net, a real-time recognition network for automated stomatal trait acquisition.

Main Methods:

• Developed SLPA-Net, a real-time recognition network for stomata localization and phenotypic analysis.
• Utilized ECANet to enhance the accuracy of stoma and aperture detection, addressing challenges with small stomata and background similarity.
• Implemented Focal EIoU Loss to mitigate issues with bounding box regression imbalance.

Main Results:

• SLPA-Net demonstrated excellent performance in detecting and identifying stomata and apertures.
• The network showed strong migration generalization and robustness in stomatal and aperture detection.
• Phenotype data acquired by SLPA-Net correlated well with manually obtained data.

Conclusions:

• SLPA-Net offers an efficient and accurate automated solution for high-throughput stomatal phenotype analysis.
• The developed network can significantly aid in breeding crops with improved water use efficiency and stress resistance.
• This approach overcomes the limitations of traditional manual measurement methods in plant phenotyping.