PhytoOracle

PhytoOracle processes multi-modal phenomics data to enable scalable, reproducible extraction of plant phenotypic traits from RGB, thermal, PSII chlorophyll fluorescence 2D images, and 3D point clouds for genotype–phenotype analyses.

Key Features:

• Modularity and Extensibility: Components are packaged as standalone containers to ensure transferability and extensibility across computing environments.
• Scalable Data Processing Pipelines: Uses open-source distributed computing frameworks to enable parallel processing on high-performance computing clusters, cloud-based systems, and local machines.
• Multi-Modal Data Fusion: Integrates RGB, thermal imaging, PSII chlorophyll fluorescence 2D images, and 3D point clouds to produce fused phenotypic data that bridge genotype and phenotype.
• Reproducibility: Implements reproducible computing frameworks to support consistent results across studies and environments.
• Efficient Processing Times: Demonstrated processing using up to 1,024 cores with 235 minutes for RGB (140.7 GB) and thermal (5.4 GB) images and 13 minutes for PSII images (86.2 GB).
• Trait Extraction Across Species: Supports phenotypic trait extraction in lettuce and sorghum.
• High Repeatability Values: Reported repeatability for Field Scanalyzer data of 0.39–0.95 for bounding area and up to 0.95 for axis-aligned bounding volume and plant height, and 0.55–0.95 for drone-derived data.

Scientific Applications:

• Genotype–Phenotype Analysis: Enables large-scale analyses of plant traits to elucidate genotype–phenotype relationships using fused multi-modal datasets.
• Crop Improvement and Agricultural Research: Processes Field Scanalyzer and drone-derived datasets to quantify traits relevant to lettuce and sorghum breeding and phenotyping studies.
• High-Throughput Phenomics: Provides scalable, reproducible processing for studies requiring RGB, thermal, PSII, and 3D point cloud data at large scale.

Methodology:

Components are containerized and orchestrated with open-source distributed computing frameworks to enable parallel processing on high-performance computing clusters, cloud-based systems, and local machines and to integrate RGB, thermal, PSII 2D images, and 3D point clouds.