Machine learning-based automatic reinforcing bar image analysis system in the internet of things

Research on the analysis of reinforcing bar images has been conducted to count reinforcing bars moving along a conveyor belt at a bar production plant. It is relatively easy to analyze images at the plant, where the environment and light sources can be tightly controlled. At construction sites, the characteristics of images vary greatly depending on the environment, time of image acquisition, and weather conditions. Therefore, a method for correctly segregating the reinforcing bar area is needed. In this paper, we propose an automatic reinforcing bar image analysis system based on machine learning. Our proposed system accurately separates the bar area from the background and counts the number of bars in the image. Compared with existing method, the proposed system performs better on detection of reinforcing bars.