Research On The Application Of Deep Learning Algorithm In The Inspection Vehicle Of Hazardous Chemicals Warehouse

Hazardous chemicals warehouse is used to store flammable,explosive and highly corrosive hazardous chemicals.Dangerous chemicals storage has very strict requirements.Some dangerous varieties are strictly prohibited to be mixed and have strict safety distance requirements.Violation of these regulations may lead to serious safety accidents.The storage safety of hazardous chemicals is a problem that needs to be solved urgently.Deep learning has emerged in various fields,but the hazardous chemicals warehouse environment is complex,and convolutional neural networks require fast and adaptable training methods.In this paper,the deep learning algorithm is applied to the inspection car vision,and two training methods of convolutional neural network are proposed to simplify the training model and improve the training speed and applicability of the model.In order to train a streamlined and fast-training network architecture and suppress the occurrence of overfitting during neural network training,a dropout method based on Poisson distribution is proposed in this paper.On the basis of making full use of the historical behavior of neurons,the neurons are screened at the fully-linked layer;the experimental results show that,while maintaining the correct rate,the loss value converges in advance,saving training time.In order to quickly extract more effective features in the data set,a convolutional neural network algorithm based on double pooling is proposed in this paper.In order to highlight the edge information in the image,this method convolves the data area with a given template to speed up the training speed of the network,so as to quickly obtain a lower loss value.This method is easy to implement and is suitable for scenes that rely on image edge information for recognition.