Cigarette Packaging Quality Inspection Based on Convolutional Neural Network

In recent years, in the process of cigarette production, there is the quality problem of cigarette package, which has become a very important factor restricting manufacturers to comprehensively improve the quality level. In order to improve the production quality of cigarettes, the quality inspection of cigarette package is very important. In this paper, an improved Bilinear-VGG16 model is proposed for cigarette carton quality detection. Firstly, images of qualified cigarette pack and defective cigarette pack are collected and enhanced to establish a data set of cigarette pack images. Secondly, the original convolutional neural network model is analyzed and improved, and a lot of training is carried out. Then, the trained model is used in the quality inspection of cigarette pack, and the defective cigarette pack is removed, and the inspection is continued if there is no defect. The final experimental results show that the improved network model in this paper has a high accuracy process for the quality detection of collected cigarette packs, reaching 96.3%, and the average detection time has also been improved. Compared with the original method, this method is more efficient than the traditional cigarette packaging quality detection method.