Open AccessResearch on Image Recognition Method of Convolutional Neural Network with Improved Computer Technology
Author(s) -
Xiaohong Li,
Xiangfeng Lv
Publication year - 2021
Publication title -
journal of physics. conference series
Language(s) - English
Resource type - Journals
SCImago Journal Rank - 0.21
H-Index - 85
eISSN - 1742-6596
pISSN - 1742-6588
DOI - 10.1088/1742-6596/1744/4/042023
Subject(s) - convolutional neural network , computer science , artificial intelligence , abstraction , image (mathematics) , pattern recognition (psychology) , computer vision , point (geometry) , mathematics , philosophy , geometry , epistemology
It has been a goal of artificial intelligence to make computers recognize objects and have the vision similar to human beings. After years of development, considerable progress has been made, but it is not satisfactory. It can be said that deep convolutional network is now the best algorithm for image recognition, which is also the reason why this paper decides to adopt deep convolutional network algorithm. The convolutional neural network USES Shared parameters between the convolutional layers, which not only reduces the required memory size, but also reduces the number of parameters to be trained and improves the performance of the algorithm. The difficulty of image recognition is hard to find a way to from the natural environment the boundary shape, texture, angular point, and image characteristics such as concept, and the image itself are highly susceptible to the influence of natural environment and changes in perspective, dimension, twist deformation, interference, light, shade, difficulties of the difference of background doping and within the class, make the computer more difficult to get about abstraction and understand the expression of natural images.