Application of Mobilenetv3 for Detecting Resin OverflowDefects in Abrasive Wheel.

In this study, an abrasive wheel defect detection system is devised, that comprising two distinct shooting methods dedicated to capturing cutting and grinding wheel images, respectively. The defect detection process consists of two stages. First, the system employs ellipse fitting to accurately capture the outline of the abrasive wheel. Second, MobileNetV3 is utilized for detecting resin defects on the surface of the abrasive wheel. Experimental results demonstrate a remarkable 99.51% detection rate for cut-off wheel resin overflow defects. Moreover, the detection rates for resin overflow defects on both the front and back surfaces of the grinding wheel reach a perfect 100%. The proposed system achieves high accuracy while maintaining efficient detection times of less than 1 s per piece for single-sided analysis.