Leaf Analyzer: A fully automated and open-source tool for high-throughput leaf trait measurement

Area of Science:

• Plant biology
• Computer vision
• Agronomy

Background:

• Accurate leaf trait measurement is crucial for plant phenotyping, agronomy, and ecology.
• Existing methods often require controlled conditions or strong contrast, limiting their applicability.

Purpose of the Study:

• Introduce Leaf Analyzer, an open-source, automated computer vision tool for high-throughput leaf trait measurement.
• Address limitations of existing methods by developing a robust background separation technique.

Main Methods:

• Developed Leaf Analyzer, a computer vision tool utilizing K-means++ clustering and a novel Leaf Background Separation (LBS) feature.
• LBS combines CIEL*a*b* and HSV color spaces for effective leaf-background distinction.
• Evaluated performance against Petiole Pro and LeafByte under various imaging conditions.

Main Results:

• Leaf Analyzer demonstrated superior accuracy and consistency compared to existing tools, especially in challenging imaging scenarios.
• The LBS feature effectively separated leaves from backgrounds across diverse lighting and color variations.
• Explored enhancements like leaf flattening and integration of texture/color features for precision.

Conclusions:

• Leaf Analyzer offers a versatile and accurate solution for high-throughput leaf trait measurement.
• Its robustness and ease of use make it valuable for diverse plant science applications in lab and field settings.
• The tool supports various applications including phenotyping, seed counting, and damage assessment.