Application of Deep Convolutional Neural Network in Computer Vision

Authors

  • Lepeng Chen College of Electricity and New Energy, Three Gorges University, Yichang 443002, Hubei, China
  • Chengjiang Wang College of Electricity and New Energy, Three Gorges University, Yichang 443002, Hubei, China

Abstract

With the rapid development of computer technology, computer vision brings a lot of convenience to people’s lives, study and work. Because computer vision is a newly developed field, there are still many problems in computer vision research. For example, it is not perfect for image segmentation, target detection, and image classification. This paper addresses computer vision, mainly to improve image segmentation and target detection in computer vision. In this paper, deep convolutional neural networks are used, which must first be constructed. Then, both the loss function of the network and the gradient magnitude are calculated. Finally, the sharing mode of the network and the target detection based on the bidirectional feature pyramid, are examined. The results show that the image segmentation and target detection using the deep convolutional neural network can make the gradient amplitude of the filler sub-branches in the basic network significantly larger than the gradient amplitude of the instance segmentation network. The sharing mode can bring about a 70% improvement in performance. Furthermore, the target detection method based on the bidirectional feature pyramid extracts more information.

Keywords: Computer Vision, Deep Convolutional Neural Network, Image Panorama Segmentation, Image Target Detection.

Cite As

L. Chen, C. Wang, "Application of Deep Convolutional Neural Network in Computer
Vision", Engineering Intelligent Systems, vol. 27 no. 4, pp. 185-192, 2019.






Downloads

Published

2019-12-01

Issue

Vol. 27 No. 4 (2019): International Journal of Engineering Intelligent Systems

Section

General Submission

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The submission of a paper implies that, if accepted for publication, it will not be published elsewhere in the same form, in any language, without the prior consent of the publisher.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.