Maintaining quality for medical equipment and pharmaceutical products is no easy task, especially when your product line includes dozens of brands sold in countries around the world.
A major manufacturer came to MLaaS in search of an automated solution for a wide range of contact lenses. From lenses designed to correct near and farsightedness, to astigmatism and presbyopia via spherical, multifocal and toric lenses, there were more than a dozen unique product-lines, each with an application-specific design, and where the range of prescription variation was met through a universe of SKUs that correspond to the individual perception.
Since the client already had machine vision systems in place, MLaaS is simply trained from large datasets containing tens of thousands of images of defective and acceptable lenses. Upon acheiving optimal classification accuracy the trained AI model is integrated into the existing system. With high levels of accuracy required, the Ground Truthing interface in MLaaS is particularly important. This allows for domain experts to inject knowledge into the image data set during the AI training process. With MLaaS defects can be identified in real time with greater accuracy and speed.