Multi-Dimensional Machine Learning Approaches for Fruit Shape Recognition and Phenotyping in Strawberry

Abstract

Background Shape is a critical element of the visual appeal of strawberry fruit and determined by both genetic and non-genetic factors. Current fruit phenotyping approaches for external characteristics in strawberry rely on the human eye to make categorical assessments. However, fruit shape is multi-dimensional, continuously variable, and not adequately described by a single quantitative variable. Morphometric approaches enable the study of complex forms but are often abstract and difficult to interpret. In this study, we developed a mathematical approach for transforming fruit shape classifications from digital images onto an ordinal scale called the principal progression of k clusters (PPKC). We use these human-recognizable shape categories to select features extracted from multiple morphometric analyses that are best fit for genome-wide and forward genetic analyses. Results We transformed images of strawberry fruit into human-recognizable categories using unsupervised machine learning, discovered four principal shape categories, and inferred progression using PPKC. We extracted 67 quantitative features from digital images of strawberries using a suite of morphometric analyses and multi-variate approaches. These analyses defined informative feature sets that effectively captured quantitative differences between shape classes. Classification accuracy ranged from 68.9 – 99.3% for the newly created, genetically correlated phenotypic variables describing a shape. Conclusions Our results demonstrated that strawberry fruit shapes could be robustly quantified, accurately classified, and empirically ordered using image analyses, machine learning, and PPKC. We generated a dictionary of quantitative traits for studying and predicting shape classes and identifying genetic factors underlying phenotypic variability for fruit shape in strawberry. The methods and approaches we applied in strawberry should apply to other fruits, vegetables, and specialty crops.

Plant phenotyping relevance

イチゴ果実画像から形状を定量化・分類する画像解析、機械学習、形態計測、PPKC手法の開発が研究の中心であり、再利用可能な表現型特徴量を生成している。

Code and data availability

The paper explicitly states that its strawberry fruit image dataset (pre-processed and processed images, extracted features CSV) is deposited on Zenodo [63] and the analysis code on GitHub [64], but no concrete public URLs or identifiers for either repository are supplied in the article blocks. The only allowed URL is,

No evidence-backed public reproduction asset is currently recorded.