Customer Story
min read

Orbisk is Using Accurate AI to Help Restaurants Reduce Food Waste Up to 70%

Orbisk is focused on limiting the amount of food waste produced in restaurants, hotels, and cafes with their AI-powered food waste monitoring solution.

Full Name*

Email Address*

Phone Number*



Thank you! Your submission has been received.
We'll get back to you as soon as possible.

In the meantime, we invite you to check out our free resources to help you grow your service business!

Free Resources
Oops! Something went wrong while submitting the form.
Orbisk is Using Accurate AI to Help Restaurants Reduce Food Waste Up to 70%Abstract background shapes
Table of Contents
Talk to an Expert

Food waste is a systemic problem that happens along every step of the food supply chain, from farm and field to fork.In the United States alone, restaurants generate an estimated 22-33 billion pounds of food waste annually. Drivers of food waste at restaurants can include over-purchasing, improper food storage, extensive menu choices, and unsuccessful employee training. These factors are only exacerbated by unpredictable supply and demand (consider the impacts of the pandemic on restaurants) and inflexible supply chains to support this fluctuation.The price of food waste is inconceivable, costing the US an estimated $218 billion per year. But on top of this staggering price tag is the cost of irreversible effects on our environment. Once food waste ends up in a landfill, it contributes to our worsening climate by releasing harmful emissions such as methane, a greenhouse gas that’s estimated to be up to 86 times more powerful than carbon dioxide.Traditionally, businesses looking to mitigate food waste have resorted to manually tracking and recording the quantity, frequency, and types of food being thrown away. But without more granular visibility into their waste streams, tackling the problem at the source remains a significant challenge.

Orbisk reduces food waste with the help of AI

The good news is that forward-thinking organizations are exploring ways that AI can help foodservice providers get better visibility into their supply chains, to mitigate the impacts of waste on the environment — and on their budgets.To learn how Orbisk uses accurate AI to help food providers make data-backed decisions to help solve the challenge of food waste, check out the video below or read on for a deep dive.

Empowering foodservice providers to make data-backed decisions

Orbisk is a company focused on limiting the amount of food waste produced in restaurants, hotels, and cafes with their AI-powered food waste monitoring solution.The proprietary Orbi device uses computer vision to help restaurants capture data about their food waste. The device is simple to use: before you dispose of food, hold it in front of the LED light and Orbi takes a picture. Orbi’s computer vision algorithms then process and categorize food on the plate to be aggregated and presented on a personalized dashboard.Orbisk was confident that better visibility into food waste would empower restaurants and other foodservice providers to make data-informed decisions about their food supply chain workflows, in order to save money — and the environment.But before they could provide their customers with accurate reports, Orbisk needed high-quality datasets to power their algorithms.

Read the Executive Summary

The challenge of accurately labeling massive datasets

To best serve their clients, Orbisk needed computer vision algorithms that could accurately identify a massive number of ingredients — vegetables, fruit, bread, condiments, sauces, desserts, meat — in short, anything that might appear on a menu.What’s more, algorithms needed to be able to identify these different foods in all their potential shapes and sizes, and from a variety of angles. For example, a tomato might appear on a plate cubed, sliced, halved, quartered, whole, as well as cooked or fresh. With all these variables, resulting datasets are necessarily massive.Orbisk knew that they could not possibly account for every ingredient and angle ahead of time; even algorithms trained on the largest datasets will eventually be faced with edge cases and new data to be trained on. They needed a labeling partner who would be able to work with them to keep their datasets up-to-date and performing well. As CEO and Co-Founder Olaf van der Veen put it:

"You can imagine the heaps of images coming in from the restaurants that we work with. Most are identified by image recognition algorithms, but for outliers and edge cases, we rely on Sama.”

Higher performing computer vision models with tight feedback loops

To start, Orbisk reached out to Sama to help them build and improve the dataset to help train their models. Sama was engaged to label hundreds of thousands of images of food. By accurately annotating images with labels such as “mashed potatoes” and “baked potatoes,” Orbisk’s AI is then able to automatically differentiate between the two. But in order to build this robust algorithm that could accurately recognize food of all shapes in sizes, and at every stage of the cooking process, they needed uncompromisingly high-quality data.To achieve this, it was important to Orbisk to remain in close communication with the team labeling their data. Sama’s annotators worked closely with the Orbisk AI team to iterate rapidly on instructions in order to reach consistent attribute tagging. Johanna Schacht, AI Team Lead at Orbisk explained:

“Sama’s agents became increasingly better at labeling our data thanks to feedback loops. This iterative way of working has made them experts on our data.”

Flexible workflows to provide clients with timely, accurate reports

Orbisk also needed a labeling partner who could help manage a constant influx of data, and scale with new clients over time. Being able to accurately identify food is the first step; the second is to analyze and aggregate that data so businesses can begin limiting food waste in real-time.Orbisk worked with Sama to integrate an API to help them send data to annotators in a streamlined way, to ensure that customers are receiving accurate reports, and in a timely manner. As Schacht explained:

“With Sama labeling images around the clock, we can send our customers reports much earlier – providing them with valuable, timely feedback.”

In short, Sama has assisted Orbisk in developing a robust solution that automatically recognizes food throughout all steps of the preparation, cooking, and consumption process. Restaurants, hotels, and cafes have access to readily available insights into which foods they are wasting most often, allowing them to better manage their budgets and waste levels.The result? Well, the proof is in the pudding: some of Orbisk’s clients have seen up to 70% reduction in their food waste. Cumulatively, this solution has helped these businesses avoid wasting 200,000 kilograms of food to date — that’s over 4,000 kilograms per location.

AI can help businesses save money and the environment

With this reduction of food waste comes significant cost savings and a measurable impact on the environment. Orbisk has the long-term goal of implementing their technology at major hotels and restaurants across the world to significantly lower the amount of waste ending up in landfills annually.If this technology is used widely, not only would it save businesses money, it would also help the planet by limiting food waste and ultimately lowering GHG emissions. By 2025, Sama and Orbisk have a combined goal to prevent 100,000,000 kilograms of food waste from ending up in landfills annually.To help achieve their lofty goals, Orbisk needed a partner whose values were aligned with their own. Sama had the added benefit of driving a proven ethical AI supply chain, having provided worker training programs to increase economic opportunity for more than 13,000 people from underserved communities. Van der Veen said it best:

“At Orbisk, we try to make the world a better place. And that’s not only true for food waste, that’s true for our entire value chain — working with Sama guarantees our ethical AI supply chain.”

Read the Executive Summary



Related Blog Articles

No items found.