GiNA

GiNA quantifies shape- and color-related phenotypic traits from conventional digital camera images of fruits, vegetables, and seeds to support horticultural trait phenotyping.

Key Features:

• Implementation: Open-source implementations are available in R and MATLAB.
• Input data: Operates on conventional digital camera images to extract morphological and color information.
• Measured traits: Extracts up to 11 morphological traits including length, width, two-dimensional area, volume, projected skin surface area, and RGB color.
• Robustness: Produces consistent measurements under varying lighting conditions and camera setups.
• Validation — morphology: Five-fold cross-validation reported correlations of 0.97 for length and 0.92 for width in cranberry fruits compared to manual measurements.
• Validation — color: Image-derived color estimates showed prediction accuracies with correlations exceeding 0.83 against total anthocyanin content (TAcy).

Scientific Applications:

• Horticultural trait phenotyping: Quantification of shape- and color-related traits in fruits, vegetables, and seeds for phenotypic characterization.
• High-throughput phenotyping: Automated extraction of multiple morphological and color traits from digital images for large-scale studies.
• Biochemical correlation: Use of image-derived RGB color metrics to predict or correlate with biochemical measures such as total anthocyanin content (TAcy).

Methodology:

Processes conventional digital camera images to extract length, width, two-dimensional area, volume, projected skin surface area, and RGB color; implemented in R and MATLAB; validated using five-fold cross-validation.