Application of Improved RCNN Algorithm in Loss Identification of Building Structures
After disaster events,how to timely and accurately assess the loss of building struc-tures has become one of the important problems in post-disaster rescue.The traditional man-ual evaluation method has the problems of time consuming and low accuracy.In order to im-prove the accuracy of structural loss identification,this study introduced the candidate region generation network based on the convolutional neural network,which was used to generate the network candidate region,classify it and perform location regression.The results show that the accuracy of Faster-RCNN in building loss identification of this research model is high-er than 90％,which is higher than the other two algorithms.In model comparative analysis,when the number of iterations is greater than 2500,the average accuracy of Faster-RCNN is higher and the fluctuation range is smaller than that of the other four models.In conclusion,the improved RCNN algorithm in this study has a good effect on building damage identifica-tion,and the results are reliable.
building structuredeep learningloss identificationfeature segmentationRCNN
NETL
NSTL
万方数据
