Machine learning-enabled computer vision for plant phenotyping: a primer on AI/ML and a case study on stomatal...

Area of Science:

• Plant Biology
• Computational Biology
• Image Analysis

Background:

• Artificial intelligence and machine learning (AI/ML) offer powerful tools for analyzing large biological image datasets.
• Estimating plant trait data from images is a key application, with stomatal traits being of significant interest.

Purpose of the Study:

• To review 39 papers on AI/ML models for estimating stomatal traits from epidermal micrographs.
• To provide plant biologists with an understanding of AI/ML capabilities and limitations in this field.
• To summarize current advancements and challenges in AI/ML-enabled stomatal phenotyping.

Main Methods:

• Systematic literature review of 39 research papers.
• Analysis of AI/ML model development and application for stomatal trait estimation.
• Evaluation of model performance across genetic, environmental, and developmental variations.

Main Results:

• Most AI/ML models achieve human-level performance for stomatal density quantification at high speeds.
• Model generalizability is often limited across diverse phenotypic variations.
• Some models predict additional traits but require substantial ground-truth data generation.

Conclusions:

• AI/ML shows great promise for accelerated stomatal phenotyping.
• Challenges remain in model applicability across phenotypic diversity.
• Future work should focus on advancing AI/ML-enabled computer vision for broader plant trait analysis.