Application Skill sets in Machine vision (and Automation)
COVID 19 has forced many industries to look for more automation and remote working. In the past 5 to 8 years, even most of the developing nations or underdeveloped nations were forced to look for automation in industries to keep pace and tackle manpower unavailability or in other words tackle lack of skilled manpower availability.
Let us look into the meaning of skill set? In general, a skill set is the knowledge, abilities, and experience needed to perform a job. Job here for our topic means automation using machine vision or assist automation using machine vision. If we look into many countries, though wanting to automate and introduce machine vision in their industry, struggle to find proper system integrators – cost effective and also job effective.
Most of the completed systems towards automation using machine vision from other countries (say developed countries) are awfully expensive (currency conversion and local customs, taxes make it worse) and does not have features to suit nativity of the problem or task or job.
In many vision applications in developing or underdeveloped countries, customers or industries or small part manufacturers look for flexibility in using the vision system for multiple tasks and re write the accuracy level of the output as they have mixed need. A label manufacturer wants to inspect labels at his will using the vision systems and not as per systems sold from abroad. They need lot of flexibility in redefining parameters according to the charges they levy on inspection and the quality consciousness of their customers.
It means most of these customers from Industries need customized vision systems with all flexibility. It is too complex to make such systems than straight forward application-based systems. These customers or industries may be looking for local companies to make such systems or integrate such systems for their use at an affordable cost.
Mostly in developing or underdeveloped countries, huge companies (who are in different fields of operations and keeps investing) do not venture into making these kinds of vision-based automation tailored to customer need as they find not worth in terms of returns and more importantly not able to locate talented, skilled engineers to work in the field. The craze of engineers coming out of institutions to join software automation or IT companies is a setback. These engineers are not given opportunity, or they do not take opportunities to harness their skill sets to become a quality vision-based automation engineers to do System integration.
Image and Vision Opportunities
Given this situation as background, I think skill set to become qualified system integrators revolve around knowledge on Cameras, Optics, Lighting, Image processing and Vision analysis techniques more importantly towards application. Knowledge on cameras, optics, lighting are now fed by the manufacturers of these equipment in different forms like conducting seminars, newsletters, videos, orientation courses etc. Image processing is part of engineering curriculum which engineering graduates’ study in their college life depending on their interest.
In system integration, application development knowledge is the essence. I find knowledge on understanding the vision analysis concepts towards application development is not taught any where or no interest shown by engineers to learn the same. Imagine a person who has profound vocabulary but does not know how to create a write up, a poem , an essay , a story etc., then his vocabulary lies waste. It is the same situation in vision application development. Engineers lack the knowledge to how and where to use the techniques of Vision algorithms already available towards the application. I have seen engineers who try pattern matching when a simple average algorithm can do presence and absence.
It is a myth with most of the upcoming engineers to use pattern matching as the only technique in Vision application development.( Introduction of Deep Learning where models created to use as a kind of finding match puts engineers off from thinking domain of using conventional machine vision algorithms towards applications.) This skill set though needs knowledge on Imaging , statistics, signal processing but is mainly about finding a path or translating these techniques towards application development. With situations demanding more flexible vision systems fine tuned to the need of industries, this skill set of approaching a vision problem with proper technique becomes important.
Skill Set Improvement
How to improve this skill set? It needs a proper mind mapping of all available resources towards an application. Can these be taught? If no, then how to acquire these skills. If yes how to do it. Let me take the case of Yes, I strongly believe a curriculum in institutions and/or in larger companies where they have skill enhancement programs to invest on labs and spread a whole sheet of application need and put the engineers to work on applications with different techniques.
Here again I am not talking about innovating a new algorithm or finding a new imaging technique or a new convolution or any sort of matrices and masks, I am highlighting about usage of tools towards application in Machine Vision. Engineers should be allowed to be trained to do application using different tools and know the consequences. If tools are available use them instead of reinventing the wheel and waste time.
Engineers have to UNDERSTAND the need of the customer in terms of Vision application and on hearing the need or reading about the need their application knowledge of using vision analysis algorithms should immediately trigger them to map the prototype on mind and then on paper. Only when an engineer knows about tools, he can understand the need in terms of vision aspect.
As a distribution and system integration company with these thoughts and our experiences in the field in India, we have developed an interactive e-book on machine vision. Objective was to provide the above said platform for engineers.
Research and Development and patents always make a country great but if the country is a developing nation or underdeveloped and aiming to grow through industries automation and machine vision are key and that means engineers with application knowledge skill sets are important. Learning or developing skill sets on conventional machine vision algorithms and using them effectively will always stand important.
Our e-book: https://www.onlsol.com/product-details/machine-vision-e-book and intro video can be seen in https://www.youtube.com/watch?v=oOh3hMG-CjQ
A write up by
Director – Online Solutions (Imaging) Pvt. Ltd.,
15A PNMK SALAI, Velachery, Chennai 600 042 INDIA
email@example.com Mob: 9840082630