Research On Recyclable Garbage Classification Algorithm Based On Deep Learning

Since the beginning of the 20th century,my country’s economy has developed rapidly,and people have moved from solving food and clothing to realizing a well-off life.However,the introduction of a large number of industries has also brought various problems.The most serious one is the problem of garbage disposal.The environment has created unprecedented pressure.With the rapid development of science and technology,deep learning technology has played a major role in military,medical,and environmental protection fields,and has promoted social development.Deep learning is a complex machine learning algorithm.The effect achieved in speech and image recognition far exceeds previous related technologies.Its ultimate goal is to enable machines to have the ability to analyze and learn like humans,able to recognize text,images and sounds.And other data.Image recognition technology is a technology that uses equipment to collect real-life images and perform feature extraction to recognize images.This technology has been widely used since its birth.The recyclable garbage classification algorithm based on deep learning studied in this paper can classify garbage efficiently and accurately,and solve the problem that people do not know how to classify garbage in daily life.According to my country’s current garbage classification standards,this article divides recyclable garbage into 5 categories.The most common Image Net data set is selected,which contains 14 million images and more than 20,000 categories,which is enough to support this experiment.The image is strengthened and noise-reduced to make the useful features of the image more obvious and easier to identify.In order to reduce the time to find the optimal solution and improve the accuracy rate,the data is normalized.According to the specific method steps of image segmentation,take out the areas that are not needed for the experiment,keep the areas that can highlight the common characteristics of a certain category of garbage,and then divide the image so that the image finally presents the sub-regions that meet the standard.Extract and separate the laboratory For the objects that are needed,this article adopts the image segmentation method of threshold segmentation and edge detection.This article selects a portable Tensor Flow deep learning framework,which has a large number of optimizable examples and trainable models.You can make full use of the model and train other content on the model to better perform migration learning.The VGG model is formed after improvement on the basis of the AlexNet model,which makes the local position information of the picture better preserved.The convolutional layer of VGG19 is 3 more layers than VGG16,and the effect is better.The convolutional layer,pooling layer,and fully connected layer in the convolutional neural network are in turn responsible for feature extraction,reducing the amount of parameters,integrating features into high-level features,and finally classifying them through the Softmax classifier.However,the actual situation is complicated,and the final effect is often not as envisaged.The use of transfer learning can improve the over-fitting phenomenon very well.This article combines Lazy Optimizer with Lookahead,which can improve the strengths and avoid weaknesses,not only improve the generalization ability and fitting speed,but also greatly improve the accuracy and stability.After testing the experimental results,it is found that the accuracy of garbage classification can be as high as 95% after using the VGG19 model and combining the optimizer.