Research On Tax Collection And Administration In ChongyangCounty Housing Lease

The continuous development of science and technology,as well as the increasing speed of current hardware upgrades,machine learning and even deep learning is an incredible development.The convolutional neural network is playing an important role in the field of computer vision.Agriculture is the ingrained industry in China.Melon and fruit agricultural products are also the necessities of our people's life,then the practical significance and development prospects of deep learning technology in the identification of melons and fruits in multiple scenes,the text is based on the common 20 kinds of melon and fruit products,using the convolutional neural network model,combined with the nighttime daytime image conversion model and the super-resolution reconstruction model,realizes the recognition of melon and fruit products in complex lighting environment.The research contents of this paper are as follows:(1)Convolutional neural network model design.Based on the accuracy and efficiency of recognition,this paper compares the AlexNet and VGG-16 networks,and then selects the former for model design according to their own needs.The layout modification and parameter optimization of the convolutional neural network based on AlexNet structure for melon and fruit agricultural products,the optimized convolutional neural network identification classification has been improved significantly and still can maintain a good recognition efficiency and achieve the light intensity.The function of recognizing melon and fruit products in the daytime environment.(2)The use of the nighttime image conversion model.This model is based on the generative adversarial network through the study of the day and night picture samples,and then generate new images to compare with the daytime original image,until the generated image and the real image are very similar,the output is to solve the problem that the illumination is not good and the dark picture recognition rate is low,after translation the dark light picture to obtain the photo with sufficient illumination,can improve the feature extraction in the image,and also help the human eye to visually distinguish the various types of fruits in the difference map.(3)The use of efficient sub-pixel convolutional neural networks.The main tool of this model is to reconstruct those fuzzy pictures,which can pass the input low-resolution image through the channel conversion and pixel arrangement processing of the efficient sub-pixel convolution layer,which is intended to "restore" those blurred pictures,so that the definition is more than the original input.Making it easier for people or machine recognition to distinguish their image content,complementing the shortcomings of the first two models.Through experiments,the unique"recognition-translation-reconstruction" process flow has been done.Through such a process flow,the problem of identification and classification of melon-fruit agricultural products under various environmental illuminations is solved.The experimental results show that After the night vision image conversion and super-resolution reconstruction,the screening rate has been significantly improved,providing certain development space for object recognition in scenes with insufficient ambient illumination.