The Research on Slope Collapse Detection Utilizing Motion Recognition Technology
In response to the issues of low monitoring efficiency,significant personal safety risks and high labor costs present in the monitoring work for river defense project slope collapses,an intelligent detection method for these collapses based on deep learning technology has been proposed.The proposed approach initially employs the attention-based lightweight U-net image segmentation model to precisely delineate the slope areas of river defense projects.Subsequently,the optical flow method is employed to recognize any movement occurring within the slope areas.Finally,a lightweight convolutional neural network is devised to classify the detected motion targets and determine whether the moving target is the slope itself.The experimental results obtained from both simulated data of river defense project collapses and actual on-site data demonstrate the effective capability of the proposed algorithm in identifying events of engineering slope collapse,thus highlighting its significant practical value.
deep learningslope collapse of river protection projectimage segmentationimage recognition
国家科技期刊平台
NSTL
万方数据
