A Survey: Image Classification Models Based on Convolutional Neural Networks

In recent years, image classification has become one of the hot spots in the field of artificial intelligence and computer vision. Image classification is a technique that simulates the human visual system to identify objects and classify images accurately, which has wide application in the fields of image segmentation, object tracking, face recognition, and etc. With the rapid development of deep learning architecture and hardware computing power, image classification methods based on convolutional neural networks have been rapidly developed. In this paper, we review architectures of convolutional neural networks, the process of image classification and put emphasis on the principles of advanced neural network models, quantization and pruning methods. We also analyze experiments based on different models and methods to compare their experimental data on different dimensions in different application situations. Additionally, we discuss some important issues and research dilemmas combined with current research progress. Finally, future research trends relating to classification methods are also provided.