Application Research of Machine Vision Platform Based on Deep Neural Network and Software Engineering

Aiming at the problem of unstable machine vision recognition accuracy caused by various kinds and inconspicuous features of workpiece defects, this paper puts forward a workpiece defect recognition algorithm based on deep learning network model and big data automatic training, and realizes its funct...

Full description

Saved in:
Bibliographic Details
Published in2023 International Conference on Networking, Informatics and Computing (ICNETIC) pp. 618 - 622
Main Authors Yu, Xiaoting, Cheng, Kun, Wang, Jing, Jiang, Nannan, Deng, Ran
Format Conference Proceeding
LanguageEnglish
Published IEEE 01.05.2023
Subjects
Artificial neural networks
Machine vision
Mathematical models
Neural network
Production
Software
Software algorithms
Software engineering
Training
Online AccessGet full text

Cover

More Information
Summary:Aiming at the problem of unstable machine vision recognition accuracy caused by various kinds and inconspicuous features of workpiece defects, this paper puts forward a workpiece defect recognition algorithm based on deep learning network model and big data automatic training, and realizes its function by software engineering. Deep neural network is to apply the established neural network mathematical model to machine vision by imitating the structure of neural network in biological individuals and its functional characteristics, that is, the ability of parallel information processing and nonlinear mapping of neural network. Using machine vision technology, the collected images are preprocessed, and a deep neural network detection and recognition model is constructed. The video event detection system based on machine vision has been widely used in various important fields, such as medical and health care, industrial production, aerospace, military and civil service industries, etc. This paper compares and analyzes a variety of mature implementation methods, and expounds a new solution based on the development direction of new hardware and the demand of large-scale popularization in actual projects, with the machine vision as the guide. The implementation process of the system is guided by software engineering, and it is implemented in a multi-modular way. It can effectively cope with many changing factors in the real environment, and has a very high practical significance.
DOI:10.1109/ICNETIC59568.2023.00133