基于AI的棉花轧工质量分级系统研究OA

In view of the problems that ginning quality inspection of domestic cotton still relied on manual sensory judgment,there was visual fatigue and experience error in detection,and the existing machine vision technology was only at the level of surface defect recognition,which could not meet the actual needs completely.According to GB 1103.1-2023 Cotton-Part 1:Saw ginned upland cotton,the AI grading system of cotton ginning quality integrating deep learning was constructed.By using the recognition model after feature extraction and optimization to the roughness of cotton appearance morphology and defects,the cotton ginning quality grading and the AI recognition of defects were realized.The results showed that the overall accuracy rate of the appearance roughness model was as high as 98.67%,accuracy rate of the detection model with fiber seed debris was 62.20%,and the recall rate was 55.60%.When the sample amount of the broken seed detection model was close to 8 000 grains,the accuracy rate of the model was reached 91.70%and the recall rate was reached 93.00%.This research realized the automatic recognition and evaluation of the defects and roughness of cotton appearance morphology,which provided a technical path for the intelligent detection of cotton ginning quality.