Convolutional neural network model-based dangerous object image classification method

Abstract

The invention discloses a convolutional neural network model-based dangerous object image classification method. The method comprises steps: a platform based on a Caffe depth learning framework comprising multiple convolutional neural network models is built on a graphics processor; a training data set and a test data set with tags are prepared, and the above data sets are used for training the above convolutional neural network models on the graphics processor; dangerous object types in China civil aviation are listed; an original image and an image after being rotated for 180-DEG in the dangerous object types are inputted to the above well-trained convolutional neural network model, a top-5 test result is obtained, and thus dangerous object image classification is realized. The method of the invention has the advantages of strong intelligence, simple method, accurate classification, quick detection speed and the like, and can be applied to automatic classification on images containing China civil aviation dangerous objects.

Description

technical field [0001] The invention belongs to the technical field of image classification, in particular to a method for classifying dangerous goods images based on a convolutional neural network model. Background technique [0002] Artificial neural network is an artificial intelligence method that is used to simulate the functions of the human brain. After experiencing a period of vigorous development at the end of the last century, it fell into a low ebb again. Inspired by new discoveries in the field of biology and neurology in the field of animal and human brain visual nerves, deep learning technology simulates the hierarchical working mode of the visual system, and builds a deep network with a hierarchical structure on the basis of artificial neural networks. The model has brought a new development direction to the artificial neural network. [0003] Convolutional neural network is a new type of artificial neural network produced by combining artificial neural netwo...

AI Technical Summary

Problems solved by technology

Therefore, for images that cannot be recognized by the convolutional neural network, it is possible to obtain correct classification results after rotating them and then classify them, thereby improving the accuracy of image classification, but no effective method has been found so far.

Images