Research and Construction of Image Classification Model Based on Deep Adaptive Network Method
Yihang Tang, Wen Zhou, Lu Tian, Rui Guo, Jiangkai Ma, Danling Lv · 2022 2nd International Conference on Computer Science, Electronic Information Engineering and Intelligent Control Technology (CEI) · 2022
Due to the limitations of convolutional neural network in the process of image classification, this paper proposes an image classification model fused with deep multi-feature transfer network based on the existing image classification models, and improves the accuracy of image classification by optimizing the convolutional neural network model. The convolutional model optimized in this paper adopts an eight-layer network structure, and the eighth layer is regarded as an adaptive layer. The loss value of the network is about 81% accurate. Finally, the deep network adaptation algorithm is used to compare the accuracy of the two pre-training models, AlexNet and ResNet, and it is obtained that the migration rate of ResNet after adaptive layer processing is significantly higher than that of AlexNet, and the accuracy rate is increased from 75% to 80%, the classification effect is significantly improved.