Quality inspection of juice boxes
Ensure undamaged straws accompany packaged juice boxes
Empowers users to quickly set up and deploy even the most challenging 2D and 3D vision applications.
Juice boxes are presented for sale in a wrapper that also includes a bent flexible straw. The only way to drink from the box is to puncture a foil-sealed hole with the straw. A missing, misplaced, or damaged straw leads to an unacceptable product.
A missing or damaged straw leaves the juice box unusable, and an unhappy customer, usually a child, with consequent lost sales.
The juice boxes have a wide variety of designs and the colors of the straws vary as well. The straw’s position and any possible damage to it can be almost anywhere. Conventional machine vision cannot reliably distinguish the straw against the complex printed background and cannot detect the wide range of defects and misplacements in the image.
Edge learning technology easily verifies the presence of an undamaged straw. It trains on small sets of images of both included undamaged straws and of the various unacceptable conditions of missing, damaged, or mispositioned straws. The classification tool learns to divide all images into either acceptable or unacceptable conditions, disregarding any background. Classification is a particularly quick process since there is no need to identify and define specific defects. Only acceptable/unacceptable decision is required.
Consumers are guaranteed that each juice box is packaged with a functional straw, minimizing meltdowns, particularly among children.