Title: Research on computer vision capture technology based on deep convolution neural network algorithm

Authors: Bo Pan

Addresses: School of Mathematics and Statistics, Lingnan Normal University, Zhanjiang City, Guangdong Province, China

Abstract: The current vehicle detection methods have some problems, such as poor recognition rate, easy to be affected by illumination and occlusion, so they need to be further improved and optimised. This paper studies the construction of target detection model based on deep convolution neural network, improves and optimises the convolution neural network by using linear discriminant analysis and constructs LDA-CNN target detection model based on the optimised convolution neural network for vehicle detection and recognition. The results show that the detection accuracy of LDA-CNN model in complex situations is 95.67, and the lowest loss value in the training process is 0.17. The above results show that the target detection model based on improved convolutional neural network can minimise the influence of illumination and occlusion, improve the detection accuracy and efficiency and has high practicability.

Keywords: convolutional neural network; computer vision; target detection; linear discriminant analysis.

DOI: 10.1504/IJWMC.2023.131321

International Journal of Wireless and Mobile Computing, 2023 Vol.24 No.3/4, pp.312 - 321

Received: 18 Apr 2022
Received in revised form: 29 Nov 2022
Accepted: 12 Dec 2022
Published online: 06 Jun 2023 *

Full-text access for editors Full-text access for subscribers Purchase this article Comment on this article