Automated machine vision inspection is everywhere in today's world of manufacturing where human resources are at a premium and difficult inspections are needed at a faster pace.  With the integration of Deep Learning into these machine vision systems how do you know what these systems are passing good parts and failing bad parts?  Is using red rabbits or production managers making fake defects the way you test your systems?  Vision Validation is a new concept for machine vision systems that you can use to test your inspection system to determine if it is still operating within good parameters.  This can be done by having a challenge set of good/bad images that can be run past the system that in theory should produce the correct results.  Not only does the vision system pass/fail correctly, how do you know someone has not changed a parameter within your program that could fail good product or pass bad product?  Vision Validation is a new way to automate and routinely challenge/monitor your machine vision systems to verify that they are working correctly. 

I will explain in depth the idea and process behind how to set this up, and how to automate it so that you have confidence and control over your machine vision inspection systems.  Recalls are very expensive and potentially life threatening so this new technology can be used to help safeguard your production processes and flag any potential issues that may arise from program changes.  Vision validation is also a great way to assist with compliance with the FDA's 21 CFR Part 11 regulations.  

There are a lot of different paths I can take with this, but in the end Vision Validation is a really cool new technology that can be used to help automate the verification process for a machine vision system and then to verify that it is within the correct tolerances.