Research on solid waste plastic bottle cognitive based on YOLOv5s and deep stochastic configuration network
Keqiong Chen, Jiaxi An, Fang Yu, Tianrui Bu · 2022
To address the problems that the efficiency and accuracy of solid waste plastic bottle detection in complex environments are difficult to meet the requirements, this paper explores a solid waste plastic bottle cognition method based on YOLOv5s and deep stochastic configuration network. First, a normalized high-quality solid waste image sample collection is constructed; second, a solid waste image depth feature space is constructed based on YOLOv5s to characterize the solid waste plastic bottle feature information at multiple levels of differentiation; finally, a solid waste plastic bottle classifier is constructed based on a deep stochastic network to obtain a fast and high-accuracy classification model. In order to verify the effectiveness of this paper, the training and testing sample space is constructed by taking solid waste image samples with Hikvision MV-CE050-30GM camera, and this paper is compared with other model methods. The experimental results show that the method in this paper has greater advantages in the comprehensive evaluation of model cognitive accuracy and computational speed, and can be effectively applied to the intelligent sorting of solid waste plastic bottles to improve the work efficiency.