Computer Vision: Why You Should Care

Computer Vision & Machine Learning Finally Bring One Irrefutable Version of the Truth

   In the early days of operations, the only way to identify manufacturing error was to be physically present during the process. This included everything from supervising rejected items at the inspection area to keeping an eye on shipping and receiving docks. Nevertheless, the issue didn’t lie in the absence of technology, but the quality and type that was being used.

 It was rare to operate with one integrated system during the early days of material requirements planning (MRP) systems. As a result, problems arose due to the inaccurate timing and reporting that resulted from the disparate systems. For years, data processing errors had become the norm, not the exception. The most telling example of this presented itself within the running cycle count program, which worked to validate perpetual inventory levels in the system of record. At the time, it was unheard of to achieve 100% accuracy. Instead, anything over 85% was considered a success.

Ironically, if you fast forward to the technology we’re working with today, you’ll find that accuracy problems still remain an issue. Contemporary systems are even more disparate today, with dozens of tools and processes being used for manufacturing.This welcomes both data inconsistency and inaccuracy caused by a number of different reasons— human error, lack of training, employee turnover, theft, system synchronization, etc.

     Fortunately, computer scientists have rallied around the notion of a single version of the truth (SVOT), where an ideal data warehouse uses a single centralized database or a distributed synchronized database to collect all data in a consistent and non-redundant form. In this way, computer vision technology providesone, irrefutable version of the truth, based on the notion that pictures do not lie. 

With computer vision, a rapid sequence of time-stamped pictures— sometimes at several images per second— capture reality. Unlike closed-circuit security cameras, that require the user to review hours of pictures, computer vision learns what to sensor and only flags critical events of interest. Even better, computer vision provides all the benefits one can expect from management by wandering around (MBWA), while simultaneously watching every area of interest 24×7. It communicates alerts in real-time and summarizes findings to facilitate continuous improvement. Computer vision is agnostic and can be applied to just about any application. Not only is its accuracy approaching 100%, it also comes at a fraction of the cost of human observers with far more efficiency 

References

Chisholm, Malcolm (December 2006), “There is No Single Version of the Truth”, Information Management Magazine, retrieved 2010-03-15

Vipul Aroh, “Single Version of the Truth – What it means for data quality?”, Verdantis, November 12, 2014,http://blogs.verdantis.com/single-version-truth-data-quality/

Author:

Anthony Tarantino, PhD

Adjunct Professor, Santa Clara University – Operations and Finance, Lean Six Sigma

Six Sigma Master Black Belt, Certified Scrum Master, CPIM (APICS), CPM (ISM)

Senior Advisor to Atollogy

Tony@atollogy.com