本文二维码信息
二维码（扫一下试试看！）
Paper Defects Classification Based on Deep Convolution Neural Network and Transfer Learning
Received:June 20, 2021  
DOI：10.11980/j.issn.0254-508X.2021.10.010
Key Words:convolution neural network （CNN）  transfer learning  paper defects classification
Fund Project:陕西省教育厅专项科研计划项目（17JK0645）；西安医学院配套基金项目（2018PT54）；西安市科技计划项目（2020KJRC0146）；国家自然科学基金计划项目（62073206）。
Author NameAffiliationPostcode
QU Yunhui Computer Teaching and Research Section， Xi’an Medical University， Xi’an， Shaanxi Province， 710021
College of Electrical and Control Engineering， Shaanxi University of Science & Technology， Xi’an， Shaanxi Province， 710021 		710021
TANG Wei College of Electrical and Control Engineering， Shaanxi University of Science & Technology， Xi’an， Shaanxi Province， 710021 710021
CHENG Shuangshuang College of Electrical and Control Engineering， Shaanxi University of Science & Technology， Xi’an， Shaanxi Province， 710021 710021
Hits: 2614
Download times: 1942
Abstract:In this study， combined with the transfer learning strategy， a small sample deep convolution neural network classifier suitable for paper disease images was designed. Firstly， the parameters of first 7 convolution layers of the VGG16 network convolution layers were frozen， and the following convolution layers was fine-tuned to complete the feature extraction of paper defects. Secondly， the fully connection layers for classification were improved to meet the needs of paper defects classification. Finally， transfer learning strategy was adopted in the training process to improve the efficiency. The results demonstrat that this method could improve the efficiency and accuracy of paper defects recognition， and further strengthen the paper defects recognition function.
View Full Text  HTML  View/Add Comment  Download reader