Computer vision-based road lane glass beads quality analysis system using ANN

This study aims to analyze the adhesion amount and refractive index of glass beads applied to lanes in order to efficiently manage the quality of lane glass beads. To this end, specialized imaging equipment is developed for the quality inspection of lane glass beads. This system utilizes computer vision to detect the glass beads and extract their feature data. An Artificial Neural Network (ANN) is then used to classify the refractive index of the glass beads. The performance of glass bead detection and refractive index classification was also evaluated using a confusion matrix. The results showed an average detection precision of 0.992, a recall of 0.972, and an F1-Score of 0.981. Additionally, the average classification accuracy was confirmed to be 0.953. This developed system is significant because it allows for the assessment of glass bead adhesion and refractive index, which are difficult for inspectors to verify directly. However, further research is needed to achieve efficient lane glass bead quality management.