Objective To improve the accuracy and efficiency of the automatic grading method for yellow peaches on the food production lineMethods Based on the yellow peach automatic grading system (machine vision and hyperspectral technology), a new method for automatic detection of yellow peach quality is proposed, which integrates an improved YOLOv11 and an improved extreme learning machine (ELM). External quality images are captured by a CMOS sensor camera, and defects are identified using the improved YOLOv11 model. The external quality is determined by the fruit shape index and color. Internal quality is detected using a hyperspectral instrument, and after feature selection, the data is input into an ELM model optimized by an improved grey wolf algorithm to detect soluble solids and hardness as internal quality indicators. The yellow peach is graded based on both external and internal qualities. The performance of the method is verified through experiments.Results The experimental method effectively detects both the internal and external qualities of yellow peaches on the food production line, with a high grading accuracy and efficiency, achieving a grading accuracy greater than 95.00% and an average grading time of less than 0.3 seconds.Conclusion By combining machine vision, hyperspectral technology, and intelligent algorithms, rapid and non-destructive detection of food quality can be achieved.