Machine Vision for External Quality Inspection of Stone Fruits: A Comprehensive Review

Machine vision technology achieves high-speed, non-contact quality control for stone fruits, yet biological characteristics such as irregular suture lines, delicate skins, and stem–calyx confusion impose specific inspection bottlenecks. This review systematically synthesizes progress in defect identification, size measurement, and comprehensive grading while deconstructing the limitations of current color and shape analysis. Analysis of industrial applications reveals that while leading commercial systems handle basic threshold-based grading, the precise detection of micro-cracks and subcutaneous bruises remains a core technical barrier. Future trends emphasize the deployment of knowledge distillation and generative adversarial networks to overcome data scarcity and computational constraints on edge devices. This synthesis establishes a theoretical framework for the next generation of intelligent, high-efficiency grading systems in the global stone fruit industry.