Advancements in machine visions for fruit sorting and grading: A bibliometric analysis, systematic review, and future research directions
Benjamin Oluwamuyiwa Olorunfemi, Nnamdi Nwulu, Oluwafemi Ayodeji Adebo, Kosmas A. Kavadias
Abstract
This research conducted a bibliometric analysis of scholarly literature on fruit sorting and grading using machine vision, identifying primary themes, sources, most-cited publications, and countries. The literature and bibliometric analysis were thoroughly evaluated to consolidate knowledge, identify research trends, and propose specific research opportunities within the context of machine vision for fruit sorting and grading. Research articles from 2011 to 2023, indexed in the main collections of the Dimensions, Web-of-science, and Scopus databases, were examined. Findings were presented quantitatively, using tables and graphs to emphasize the key performance factors for article writing and citation. Upon applying inclusion and exclusion criteria, 129 out of 1,812 discovered articles were included for examination, while 1,683 studies were excluded due to non-compliance with the requirements and duplicates. Thirty-four (34) case study publications on machine vision applications for fruit sorting and grading were comprehensively examined to identify the adopted methodologies and future research opportunities. Covered methodologies include fruit varieties, data volumes, data collection, classification methods, and accuracy metrics. The study's findings indicate a significant increase in deep learning applications for fruit recognition in the recent five years (2019-2023), with excellent results achieved either by utilizing new models or with pre-trained networks for transfer learning. The research also identifies gaps and future directions for machine vision in fruit sorting and grading, such as enhancing system robustness, scalability, and adaptability, integrating multiple sensors and technological methods, and developing evaluation and comparison standards and criteria. The paper concludes that machine vision holds promise as a potent tool for fruit quality assessment, but further research and development are needed to address existing challenges and meet the growing demands of the fruit industry.