Applications of Deep Neural Networks in Image Classification

Abstract:
Image classification has always been a focus task in computer vision. Due to its great feature learning ability, deep learning has become the first choice to solve image classification. With the rapid development of deep learning in the past decade, image classification using deep learning has achieved tremendous progress. However, most existing reviews focus on the methods based on CNN, while the summary of relevant research progress in recent years is not much. This is not friendly for scholars who want to have a more comprehensive understanding of image classification. Therefore, this paper summarizes the task of image classification. In addition to the mainstream methods, the relatively novel and prominent methods and related research work in recent years are also introduced. Finally, there are some analyses of the challenges of image classification.