基于NCNN的特种设备原始记录便捷化采集方法的设计与应用

  • 打印
  • 收藏
收藏成功


打开文本图片集

摘要:文中通过分析特种设备原始记录采集过程中遇到的问题,提出一种基于CNN卷积神经网络和NCNN高性能神经网络前向计算框架的图像文字识别方法的应用,通过对识别的结果进行近义词分析、数据分类及数值规约保证了数据的准确性及识别结果的高可用性,从而有效提高检验人员现场采集原始记录的效率。

关键词:卷积神经网络;高性能神经网络前向计算框架;近义词分析;原始记录

Design and Application of an NCNN-Based Efficient Collection Method for Original Records of Special Equipment

ZOU Shanqing

(Fujian Special Equipment Inspection and Research Institute, Fuzhou 350008, Fujian, China)

Abstract: By analyzing the problems existing in the collection process of original records of special equipment, it proposes the application of an image text recognition method based on CNN convolutional neural network and NCNN high-performance neural network forward computation framework. This approach ensures data accuracy and recognition result availability through the analysis of semantic proximity, data classification, and numerical statute validation. The method effectively improves the efficiency of inspectors' on-site collection of original records, enhancing overall operational efficiency.

Key Words: Convolutional neural network; High-performance neural network forward computation; Synonyms; Original record

0 引言

近年来,随着我国经济持续发展,特种设备的数量与种类也在日益增长,人机比矛盾日益突出,因此如何在传统检验过程中应用新技术来提高检验检测的效率、提升检验检测的质量,一直是各检验机构努力的方向。(剩余5050字)

monitor