A positioning and identification method of cable tunnel stagnant water based on ZF-Faster RCNNs

• Main Authors: JIANG LIN, LIU DANBING, YUAN YONGYI, QIU WEI, HUANG SHUNTAO, ZHU WUZHOU, PEI XINGYU, LIAO YANQUN, CUI JIANGJING
• Format: Patent
• Language: Chinese; English
• Published: 11.12.2018
• 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 relates to the technical field of a computer image processing method, more particularly, to a positioning and identification method of cable tunnel stagnant water based on ZF-Faster RCNNs. The method comprises the steps of generating a candidate area, extracting features, judging categories and refining positions. An RPN network is adopted to extract suggestion of water-trapped location area, and then a Faster R-CNN detection network is utilized to perform image recognition, SVM classification in the suggestion box, and finally the position of the suggestion box is refined. A conventional Softmax classification method of a Faster R-CNN detection network is replaced by SVM, thus avoiding the approximate result of probability calculation caused by the selection of regularization parameters and enhancing the confidence of classification. The invention can realize the real-time monitoring of the water accumulation condition in the cable tunnel, and can realize the rapid identification and positionin