Machine Vision and the Automation of Food Inspection
By the end of 2018, the U.S. food and beverage industry will have produced enough goods to generate $16.2 billion in revenue. To put it in perspective, that number trails just behind China, the world’s No. 1 food and beverage producer by revenue, and the most populous country in the world with nearly 1.4 billion people.
It takes a complex dance of processing, production, packaging, and distribution to keep America’s behemoth food and beverage industry operating smoothly. Things don’t always go according to plan, and producers may find themselves facing product recalls, foodborne outbreaks, and even legal action at the hands of consumers. Industry 4.0 offers a smart and efficient method for ensuring that the highest level of food inspection and safety standards is met — with the help of machine vision.
Machine Vision Explained
Only recently considered a widely applicable tool for the food production industry, machine vision actually was first utilized for food quality inspection in the 1980s. These early iterations of machine vision consisted of little more than a camera taking photos and were used to inspect muffin production lines to ensure that oversized products didn’t make their way into machines. Crude though the process was, it proved effective.
Today’s machine vision systems are far more sophisticated. The Automated Imaging Association (AIA) classifies machine vision as any combination of hardware and image analysis software that assists in the functionality of a device by capturing images. The machine vision systems of today usually have more or less complex AI under the hood that analyzes patterns and extracts data from the objects within its line of sight. The data obtained is then compared with any existing data drawn from the system’s database — which is managed by humans, not automation. Once the data are reviewed, the system produces a conclusion about the item captured.
The entire process, from start to finish, takes less than a second. But, in just this short amount of time, the system gleans myriad useful insights about the item in question. Data about a food’s color, ripeness, degree of spoilage, and internal temperature is obtained in the blink of an eye. It’s even possible to obtain information that human eyes are unable to detect — such as the internal makeup or ingredients in a food — which machine vision obtains by analyzing the item using different light wavelengths. Packaging flaws are also detected, preventing materials waste, mislabeling, and costly food recalls.
According to the U.S. Food and Drug Administration and the U.S. Department of Agriculture’s Food Safety and Inspection Service there were 456 food recalls in 2017. These include multiple food safety violations/recalls for the same product. Undeclared food allergens (especially in dairy) topped the list of reasons for a recall. Listeria was the second most common reason for a recall, typically affecting popular breakfast food. There were also 24 Salmonella-based recalls, and 2.4 million pounds of ready-to-eat breaded chicken products which contained undeclared milk content, putting those with dairy allergies at risk.
While the total number of recalls was down from 2015 and 2016 — the latter of which had an unusually high number of outbreaks, resulting in stricter food safety regulations — recalls remain a problem area in food production. Indeed, as an industry, food production lags far behind those in other line industries when it comes to utilizing automation, despite the fact that increased use of machine vision could significantly improve health and safety standards.
The food production industry walks a fine line. It must maintain a high level of efficiency while producing food products that meet both consumer demand, as well as federal health and safety regulations. Additionally, producers need to weigh the cost of traditional low margin operations with the demand for greater safety and quality in the food production value line.
Implementing machine vision may require a higher initial investment than what many producers are used to, but the increase in efficiency, reduction in costly errors, oversights, and recalls (some of which may lead to even costlier litigation) can more than make up for the expense in a relatively short amount of time.
There is also the consumer cost to consider. With the health and safety of consumers at stake, the food inspection process must have a zero tolerance policy for mistakes. It takes only one misstep for a brand or product to lose the trust of its customers. An improperly labeled food item or an overlooked sign of infection in meat or poultry could cost lives. This year alone, the FDA has already tied 44 deaths to a Salmonella outbreak related to the dietary supplements containing kratom, a tropical tree native to Southeast Asia. Had machine vision or other automated inspection processes been in place at the production plant where these supplements are made, these deaths might have been prevented.
Applications for Machine Vision in Food Inspection
When it comes to food product safety, machine vision provides an unparalleled level of quality, precision, and efficiency. From raw ingredient detection to identifying whether a food item has been over- or undercooked, automated visual food inspection is able to capture even the most minute details of a food item.
However, the food inspection process doesn't end with a review of the food item alone. If the packaging is compromised in any way, the food is likely to be degraded as well. Machine vision can spot packaging defects, identify and correct misaligned labels, and even pick out erroneously labeled packing due to human or machine error.
If a food product is part of an intermediary process or is itself an intermediary product, machine vision can help track and trace all related ingredients (and their related data) throughout the production process, up to the finished product. This is an especially crucial step for food producers who receive food components from other producers, as ensuring a high level of quality and safety at each step in the supply chain becomes increasingly complex with each additional input.
Offering strong quality control and more efficient production processes, automated visual food inspection is well-positioned to innovate the food production industry. Machine vision is more than just the latest technological advancement in food inspection, it’s also an opportunity for the food production industry to raise the bar on health and safety standards — something we can all get behind.
Information about the author
Yaroslav Kuflinski is an AI/ML Observer at Iflexion. He has profound experience in IT and keeps up to date on the latest AI/ML research. Yaroslav focuses on AI and ML as tools to solve complex business problems and maximize operations.