This website uses cookies primarily for visitor analytics. Certain pages will ask you to fill in contact details to receive additional information. On these pages you have the option of having the site log your details for future visits. Indicating you want the site to remember your details will place a cookie on your device. To view our full cookie policy, please click here. You can also view it at any time by going to our Contact Us page.

Close

Having the vision to overcome inspection challenges

22 October 2021

Dr. John Dunlop discusses advances in product inspection technology which could offer huge benefits to food producers. 

From production through to packaging, food processing and manufacturing is becoming ever more automated as vision systems and artificial  intelligence (AI) has become more widely accepted.

However, despite the food industry having many opportunities for vision automation, there still seems to be a slow take-up, or understanding about how subjective tasks such as quality inspection can be automated. 

Many still think human inspection is needed for tasks such as quality inspection because the process is based on visual appearance – it’s a very subjective measure. 

Traditionally, manual operators have been provided with an image guide to distinguish what a ‘pass’ or ‘fail’ looks like. Over time, an operator will begin to learn and automatically recognise the required standards. But how can you teach an automated vision system to do the same? This is where AI, specifically Deep Learning (DL) excels.

DL is a different way of programming a system. It is based on experiential learning rather than a set of parameters. So, as you would with a manual operator, you show a DL system both good and bad example images and it will learn the difference between the two. 

When you get a DL-based result, it is referred to as classification. It will provide a percentage - e.g. ‘this is a 90% match for a good result compared to the photo examples’. This process is similar to humans acknowledging when food is cooked. We don’t need to refer to an example image every time we cook lasagna at home thanks to our learned experience. 

Classification is the significant change that so many industries, including food processing, are trying to achieve. It’s not necessarily about taking people out of the process but more a focus on adding value and upskilling operators. The addition of automation allows for an increase in repeatability and reliability within processes that is  unachievable by human operators. 

What can DL do?
As well as inspections such as colour, DL can automate inspection processes such as texture, shape, finish and even things like shine. Many products – such as pizza – will need multiple inspections –  Are the ingredients on top of the pizza? Is the pizza baked? Does it have the right quantity of toppings? Are the toppings spread evenly across the pizza? This can all be completed using DL technology.

Additionally, many people don’t realise that DL systems are also capable of identifying text – such as ‘happy birthday’ on a cake for example. If an operator is reading ‘happy birthday’ 1,000 times per day, it is extremely likely that they will become blind to errors. Whereas an DL system will examine the style of the writing and the patterns in the icing such as the knots where the writing joins. This is fool-proof and prevents missed errors. 

In 10 years, I predict it will be very unusual to see an operator undertaking manually quality inspections in food production environments. Automated inspection could quickly solve the difficult human-resource issues that companies are facing today such as finding suitable staff, staff shortages and the cost of hiring staff. These are the issues that companies are faced with now and Artificial Intelligence will play a big part in overcoming these. 

Companies are constantly looking to improve productivity and increase output but challenges such as staff shortages will prevent this. Artificial Intelligence will be what pushes these companies further – there is no point in having an automated production line or conveyor system if your final quality inspection will still be carried out manually by staff.

As well as staff limitations, manual inspections can be subjected to human error and fatigue. The only solution is to automate. 

Dr. John Dunlop is CTO and Founder – Bytronic Vision Automation.


Contact Details and Archive...

Print this page | E-mail this page