A Systematic Review and Comparative Meta-analysis of Non-destructive Fruit Maturity Detection Techniques

Abstract

The global fruit industry is growing rapidly due to increased awareness of the health benefits associated with fruit consumption. Fruit maturity detection plays a crucial role in fruit logistics and maintenance, enabling farmers and fruit industries to grade fruits and develop sustainable policies for enhanced profitability and service quality. Non-destructive fruit maturity detection methods have gained significant attention, especially with advancements in machine vision and spectroscopic techniques. This systematic review provides a concise overview of the techniques and algorithms used in fruit quality grading by farmers and industries. The study reviewed 63 full-text articles published between 2012 and 2023 along with their bibliometric analysis. Qualitative analysis revealed that researchers from various disciplines contributed to this field, with techniques falling into 3 categories: machine vision (mathematical modelling or deep learning), spectroscopy and other miscellaneous approaches. There was a high level of diversity among these categories, as indicated by an I-square value of 88.37% in the heterogeneity analysis. Meta-analysis, using odds ratios as the effect measure, established the relationship between techniques and their accuracy. Machine vision showed a positive correlation with accuracy across different categories. Additionally, Egger's and Begg's tests were used to assess publication bias and no strong evidence of its occurrence was found. This study offers valuable insights into the advantages and limitations of various fruit maturity detection techniques. For employing statistical and meta-analytical methods, key factors such as accuracy and sample size have been considered. These findings will aid in the development of effective strategies for fruit quality assessment.