Deep Learning Model for Image Classification in Machine Vision

Authors

  • Xiaojun Zhang School of Artificial Intelligence, Guangzhou Huashang College, Guangzhou 511300, Guangdong, China

Abstract

In this study, a deep-learning model based on Convolutional Neural Network (CNN) was applied in order to automatically extract image features and improve the accuracy and robustness of classification. A CNN model with multi-layer convolution and pooling structure was constructed, and the ReLU activation function was utilized to improve the ability of nonlinear expression. Multi-category classification was achieved through the Softmax layer. The training dataset was expanded through data augmentation technology to decrease the risk of overfitting and increase the adaptability of the model in complex scenes. A transfer learning strategy was adopted to use the model pre-trained on the large-scale dataset, ImageNet, to decrease the dependence on labeled data, and the model weights were optimized by fine-tuning. Using depthwise separable convolution, a lightweight network
structure was implemented to optimize the resource constraints of edge devices. Experimental results showed that the average inference latency of the model was 50.76ms and the average throughput was 20.2 FPS. The model still maintained high classification performance in a low computing environment.

Keywords: Deep Learning; Convolutional Neural Network; Data Augmentation; Transfer Learning; Lightweight Model

Cite As

X. Zhang, "Deep Learning Model for Image Classification in Machine Vision", Engineering Intelligent Systems,
vol. 33 no. 4, pp. 405-415, 2025.


 

Downloads

Published

2025-07-01

Issue

Vol. 33 No. 4 (2025): International Journal of Engineering Intelligent Systems

Section

Article

License

The submission of a paper implies that, if accepted for publication, it will not be published elsewhere in the same form, in any language, without the prior consent of the publisher. Before publication, authors are requested to assign copyright to CRL Publishing Ltd. This allows CRL to sanction photocopying, and to authorize the reprinting of issues or volumes according to demand. Authors' traditional rights will not be jeopardized by assigning Copyright in this way, as they retain the right to reuse the material following publication, and to veto third-party publication.