Research on Image Classification Method Based on DCNN

Chao Ma, Shuo Xu, Xianyong Yi, Linyi Li, Chenglong Yu

Research output: Chapter in Book/Report/Conference proceedingConference contribution

9 Scopus citations


Image classification is a kind of image processing technology, which can recognize different things by the feature information given by pictures. With the rapid development of science and technology and people's higher and higher demand for quality of life, image automatic classification technology has been applied to various fields of development. When we classify the image, the traditional image classification method can not accurately grasp the internal relationship between the recognition objects, and the traditional method also has the limitation of the recognition object's feature expression because of the too high characteristic dimension of the data, so the experimental results are not ideal. In view of the above content, this paper proposes an image detection method based on convolutional neural network. The experimental algorithm mainly refers to deep learning and convolutional neural network. Different from the traditional image classification methods, the deep convolution neural network model can be used for feature learning and image classification at the same time. By improving the structure of each part of the experiment and optimizing the convolution neural network model, the over fitting phenomenon can be prevented, and then the accuracy of image detection can be improved. The experiment on cifar-10 database shows that the improved deep learning model of this method has achieved effective results in image detection.
Original languageEnglish (US)
Title of host publication2020 International Conference on Computer Engineering and Application (ICCEA)
Number of pages4
ISBN (Print)9781728159041
StatePublished - May 30 2020

Bibliographical note

KAUST Repository Item: Exported on 2020-10-01


Dive into the research topics of 'Research on Image Classification Method Based on DCNN'. Together they form a unique fingerprint.

Cite this