• Tim Hammerich

Fighting Food Waste with Hyperspectral Technology



Hyperspectral imaging was originally developed for NASA.


The technology collects and processes how a particular object reflects light. Every object has its own unique spectral fingerprint or signature. This allows us to identify differences between objects, even when they are not detectable by the human eye.


Abi Ramanan had already started two companies in the food industry when she learned about how this technology was being applied to satellite imagery. She discovered the hyperspectral sensors were advancing to become smaller and more affordable. This lead her to think about ways she could apply the technology to the food system.


From there, Impact Vision was born. The software company installs sensors in food processing and distribution environments. These sensors, when combined with their software that utilizes machine learning and computer vision, provide real-time insights into factors like food quality and freshness.


Why?


“The idea was to equip food supply chains with digital tools” says Ramanan, “so that you can facilitate a shift towards a system where 100% of products are tested as opposed to just 2–4% today.”


In commercial food processing and distribution facilities, individual samples are usually taken and tested for quality, safety, etc. These samples are meant to be representative of what’s being sent out, but inevitably there is variation that cannot be full accounted for when only sampling 4% of less of the product.


With hyperspectral sensors installed in these facilities, the company can collect data on 100% of the products in real time. This allows them to sort production based on factors such as ripeness, which can ultimately decrease waste. Basically, it allows us to collect the chemical information of a large volume of products the moment they pass by on a conveyer.


“There actually is a lot of information that exists in the world, it’s just that our eyes don’t have the capacity to see gases…or understand things like tenderness or ripeness, whereas these cameras when packaged together with ground-truthed data, can allow us to see so much more information, and it’s not invasive and it’s real time. So this really means they can be used as tools to really help improve the efficiency in supply chains.” — Abi Ramanan


How Does This Reduce Food Waste


You probably already know that 30% — 40% of all calories produced ultimately get thrown away. This is an enormous waste of resources, and a very complicated problem to solve because waste happens for a variety of different reasons all along the value chain.


“One thing that we really need to improve is the efficiency of distribution of food through the supply chain” says Abi. “Today, a third of all the food produced is wasted, and in developed countries, about 50% of this happens during processing in the supply chain…One of the reasons for this is that food supply chains have not been equipped with tools to improve efficiency. If you’re testing one in a thousand or one in two thousand samples and then applying a statistical model, you’re going to get really uneven in quality (which means ) a lot of waste at the restaurant level, a lot of waste at the retail level, etc. It’s just because you can’t accurately predict the quality or the shelf life. So what hyperspectral imaging and other techniques allow you to do is to understand the chemical composition of every single product, whether it’s passing by on a conveyer belt for example, so every avocado can be sort of based on it…Then sorted into buckets with like content. These buckets can then be ripened together, generating a much more uniform end product.”


Where’s the Beef?


“Take the example of beef” says Abi. “We worked on a pilot with a retailer in the U.S. Meat that has a pH of above 5.9 should not be vacuum packed and sold as steak. It will almost certainly be rejected by the consumer…However, if you are able to access this information higher up in the supply chain, at the distribution center for example, then than meat can be turned into ground beef so it doesn’t have to be wasted. That’s the short term solution, but longer term a high pH indicates a problem with the supplier so you know have objective data to back up any claims that you make. And similarly, retailers have a lot of power in the supply chain and can reject whole batches of products based on samples they’ve tested. But if a producer has a guarantee that every single tomato or apple has been objectively classified and guaranteed to be of a certain quality, it just enables much more rigor in terms of the supply chain. Those examples show how you can reduce waste and how you can command a premium for better consistency of the end product.”


Beef is a great example of a product that is a prime candidate for Impact Vision’s technology: a high-value product with a short shelf life for which consumers are willing to pay more for quality. Abi also points out the resources required to grow meat as another reason to make waste of this product a priority.


Beyond quality and freshness, Impact Vision plans to use their technology to identify foreign objects such as cardboard and plastic that make their way into the food supply. These objects can often go undetected by an inspector and are not picked up by magnets like other foreign materials. The company plans to launch a pilot for this use case with a sugar company in the near future.


Fighting Food Fraud


Another problem hyperspectral imaging could tackle is food fraud, which is selling one product but delivering a cheaper substitute without the customer being able to tell the difference.


It may seem like these are isolated cases, but there is a lot of evidence that the problem is prevalent in many areas of the food supply. Abi mentions examples such as olive oil and wild-caught fish.


The idea is that hyperspectral technology could be able to identify fraudulent food in the supply chain if the spectral fingerprint is different.


Abi also envisions a future where a consumer can attach a hyperspectral sensor on their smart phone in order to determine the freshness, integrity, and quality of food in the grocery store. Pretty cool stuff!