Research on Dry Beach Length Measurement Based on Improved Mask R-CNN Algorithm

Abstract

The length of dry beach is one of the important parameters reflecting the stability of tailings dam, So dry beach length is an important aspect of tailings pond safety monitoring. In order to measure the dry beach length accurately, a deep learning-based dry beach length measurement method is proposed. The method is divided into 4 parts: (1) Install a camera to obtain tailings pond images; (2) Use the improved Mask R-CNN algorithm to train a network model that can identify the waterline position; (3). Perform contour extraction, obtain pixel coordinates and filter out the minimum pixel ordinate; (4). According to the monocular visual ranging model, the transformation formula between image coordinate system and dry beach coordinate system is obtained, and the length of dry beach is obtained by putting the minimum pixel coordinate into the transformation formula. Experimental results show that the model can accurately measure dry beach length with an error of less than 2.7%.