Carrot grading method and system based on machine vision and weight distribution

The invention provides a carrot grading method and system based on machine vision and weight distribution, and relates to the technical field of agricultural product appearance quality detection and grading. Comprises: acquiring a carrot image; the method comprises the following steps: automatically detecting carrot surface defects by adopting a technology combining deep learning and image processing to obtain carrot sizes, defect types and characteristic attribute values of each defect; judging whether the carrots are out-of-class products or not based on the defect types; distributing a weight value to the defect characteristic attribute value based on an entropy weight method, and calculating a carrot appearance quality score according to the weight value and the defect characteristic attribute value; and determining the carrot quality grade based on the carrot size and the carrot appearance mass fraction. According to the method, the carrot surface defects are automatically detected by combining deep lear