ARADEEPOPSIS

ARADEEPOPSIS performs unsupervised semantic segmentation of top-view plant images to quantify leaf tissue states (healthy, anthocyanin-rich, and senescent) for high-throughput quantitative plant phenotyping.

Key Features:

• High-throughput parallelized processing: Enables processing of large image datasets using parallelized operations for scalable phenotypic measurement extraction.
• Unsupervised semantic segmentation: Classifies leaf tissue into three categories—healthy, anthocyanin-rich, and senescent—using semantic segmentation on top-view images.
• Robustness to color and background variation: Maintains segmentation performance under variations in plant color and background conditions that confound traditional color-based methods.
• Quantitative phenotyping across conditions: Provides measurements applicable across developmental stages, mutants with aberrant leaf colors, and plants under stress.
• GWAS-compatible outputs: Produces quantitative traits that have been used in genome-wide association analyses to identify loci related to anthocyanin production and early necrosis.
• Cross-species image applicability: Operates on images from diverse origins, including distantly related Brassicaceae family members.

Scientific Applications:

• Linking phenotype to genotype: Extracts detailed phenotypic measurements that support association of observable traits with underlying genotypes.
• Genome-wide association studies (GWAS): Has been applied to GWAS, including analyses of 210 natural Arabidopsis thaliana accessions, to identify loci related to anthocyanin and early necrosis.
• Comparative phenotyping across species and treatments: Facilitates quantitative comparison of phenotypes across species, mutants, developmental stages, and stress conditions.

Methodology:

Applies semantic segmentation to top-view plant images to perform unsupervised classification of leaf tissues into healthy, anthocyanin-rich, and senescent classes, and processes images from diverse origins and quality levels while remaining robust to plant color and background variation.