Kitro has developed a tool that allows managers of hospitality businesses to analyse their avoidable food waste. Kitchen staff can learn a lot from their food waste in terms of environmental and economic sustainability.
Picture: Kitro
Food waste occurs at canteens, luxury restaurants and local snack bars alike. In total, around one third of all food ends up in the rubbish, is lost or is wasted. This is not limited to the hospitality sector: It happens along the entire value chain. In Switzerland alone, waste amounts to 2.8 million tonnes per year. As the production, processing and distribution of food account for almost a third of all greenhouse gas emissions, the potential savings in terms of environmental impact are huge. Reducing avoidable food waste also offers enormous potential cost savings, especially for restaurant businesses that sell a lot of individual portions.
This is where the start-up Kitro comes in. Anastasia Hofmann and Naomi MacKenzie both graduated from EHL Hospitality Business School in Lausanne, then founded a company that developed a system to keep track of food waste. While the retail segment already keeps records of its food waste, very few restaurants or canteens do the same. However, waste is a particularly good source of information about which processes could be made more resource-efficient, which dishes could be served in smaller portions, or where it might be possible to improve quality.
To analyse food waste, Kitro’s device, which comprises an industrial weighing scale, a camera and a mini-computer, is set up in kitchens and the existing bin is placed on it. Whenever something is discarded, the scale detects a change in weight. That triggers the camera, which then photographs the waste. This photo is uploaded to a server. There, the image is first checked to see whether it contains any errors, e.g. if the camera lens is steamed up, if the view of the rubbish bin is obstructed by a plate, or if the bin is closed or outside the frame. In the second step, the background is cut out so the image can be compared with previous images. This process determines what has been added. Philipp Andermatt, Head of Machine Learning at Kitro, explains that the algorithms used are similar to those that compare satellite images to track urban development. In the final step, the remaining image content is categorised by means of machine learning.
For all these steps, the software uses tools from the research field ‘computer vision’. Situated at the intersection between image processing and artificial intelligence, computer vision is about striving to teach computers to see meaningfully, thus enabling them to interrelate images that have similar content or to tag them with keywords. Computers are trained to identify relationships in images and to recognise, for example, whether root vegetables, mashed potatoes or pieces of stale bread end up in the rubbish bin. As the appearance of many foods changes significantly during processing, exact classification is sometimes very difficult. The question that then arises for Kitro is whether a more precise breakdown is of interest to the hospitality business in question. Often, a categorisation that groups various foods in classes without distinguishing between very similar-looking ones (so without determining whether parsley root or parsnip has ended up in the waste) is helpful in itself. Mixed waste is technically challenging: The software then has to estimate which of the foods contained in the image account for what share of the weight.
The results are summarised on a dashboard to help kitchen staff optimise their procedures. If certain dishes entail more return flow than others, this indicates that quantity or quality could be adjusted. Suppliers’ price lists can be fed into the system as well. Thus, the dashboard can also provide an estimate of the waste’s value in each instance. On average, according to Patrick Hoffmann, avoidable food waste in restaurants is worth seven Swiss francs per kilo. This means that in avoidable food waste, a restaurant has the potential to save 2 to 8 percent of the goods used and about 150 Swiss francs per business day. For this reason, Kitro’s team of customer success managers also helps the kitchen management to develop and implement suitable measures to curb food waste.
