Steel Bars Counting and Center Localization Method Based on Convolutional Neural Network and Distance Clusting

Automated steel bars counting and center localization can greatly improve steel bar production efficiency. Due to the complex industrial environment, traditional methods are difficult to achieve accurate steel bars counting. Convolutional neural network (CNN), which has great capability to deal with complex tasks in challenging environments, is applied in this work. In this paper, a framework called CNN-DC is proposed to achieve automated steel bar counting and center localization simultaneously. The proposed method first detects the candidate center points with a deep CNN. Then an effective clustering algorithm named as Distance Clustering (DC) is developed to cluster the candidate center points and locate the true centers of steel bars. The proposed method can achieve 99.26% accuracy for steel bar counting and 4.11% center offset for center localization on the established steel bar dataset, which can perform well on automated steel bar counting and center localization.