Equipment switch state identification method based on improved Faster RCNN

• Main Authors: JI JUNJIE, ZHU LIANG, ZHANG BINGBING, LI HAIBO, SONG XUFENG, WU MIN, XU WEI, YAO HANGYU, LU TING, JIANG JIAJIE, SHEN TAO
• Format: Patent
• Language: Chinese; English
• Published: 18.02.2022
• Subjects: CALCULATING; COMPUTER SYSTEMS BASED ON SPECIFIC COMPUTATIONAL MODELS; COMPUTING; COUNTING; HANDLING RECORD CARRIERS; PHYSICS; PRESENTATION OF DATA; RECOGNITION OF DATA; RECORD CARRIERS

The invention discloses an equipment switch state identification method based on an improved Faster RCNN, which is suitable for equipment switch state image identification by improving an original Faster RCNN model in terms of network structure, parameter setting and algorithm. The method mainly comprises the following key steps: image data acquisition, transmission and storage, equipment switch state classification, switch state image data set buidling, designing of the size of a candidate box, switch recognition model building, switch recognition model training and switch recognition model testing. Through the method, the on-off state of the equipment can be accurately identified, so that an equipment operator is assisted to carry out daily operation and maintenance work, the accuracy and reliability of the operation and maintenance process are improved, and safe and continuous operation of the equipment is guaranteed. 本发明公开了一种基于改进Faster RCNN的设备开关状态识别方法，通过对原始Faster RCNN模型在网络结构、参数设置及算法上进行改进，使其适用于设备开关状态图像识别，主