Identify problems with the bristles in toothbrush heads
Graphical programming environment for deep learning-based industrial image analysis
To manufacture a toothbrush head, bunches of nylon bristles or filaments are bent in half, inserted into holes molded in the head, and then held in place by metal staple rings. Depending on the design and intended type of brushing, bristles come in a wide range of diameters, cross-sections, and colors. After all the bristle bunches are placed, toothbrush head inspection must ensure proper placement, density, and color before it is trimmed, and the bristle ends ground into one of several possible shapes.
As with many consumer items, the range of possibilities for bristle cross-section, density, and color in toothbrush heads has grown significantly. Since toothbrushes go into the mouth, they must be free of even tiny defects. Defects are so unpredictable, and their possible range so wide, that it is difficult to program conventional machine vision to detect them all.
AI-based technology is trained using a sample set of images of acceptable heads for any given model of toothbrush, with its combination of bristle shapes, densities, and colors. During toothbrush bristle inspection the defect detection tool can then note any anomalies in the toothbrush head being inspected and reject it before it goes any farther in processing and packaging. Training on new designs requires only a new set of training images and takes only a few minutes.