Analysis of Dynamic Movement of Elevator Doors Based on Semantic Segmentation

Abstract

An analysis method for the movement state of an elevator door is proposed in this paper. Firstly, we load the monitoring videos which record the movement of the elevator door. Then we label the position of the elevator door in the video and use it as a data set for training the semantic segmentation network. Next initialize the image input layer, downsampling network, upsampling network, and pixel classification layer in the semantic segmentation network, and stack all layers to complete the creation of the semantic segmentation network. Finally, after identifying the elevator door position in the video by semantic segmentation, process the identified images using image erosion and edge detection operators and estimate the distance between the elevator doors. The method for estimating the distance proposed in this paper has strong adaptability and low cost. As a result of experiments, the accuracy of method which is proposed in this paper has reached 97.7%.