This AI tool can measure calories based on a photo

SnapCalories tech is made possible by Roboflow, which has developed advanced computer vision systems.

Personal health startup SnapCalorie has raised $2 million in seed funding from Y Combinator.

The start-up in Great Falls, USA has developed artificial intelligence technology to measure the calories in each meal based on a single photo.

Wade Norris, a former Google AI engineer, Scott Baron, a former Raytheon and MathWorks engineer, and a few others in a Google moonshot project called X are co-founders of SnapCalories, which previously raised $125,000 in a pre-seed round from undisclosed investors.

SnapCalorie claims its app, which allows users to take a photo of any meal and get an accurate calorie count, is more accurate than a trained nutritionist.

At the heart of this technology is an algorithm described in an academic paper co-authored by Norris. “Introducing Nutrition5k, a new dataset consisting of 5,000 different real foods with corresponding video streams, depth images, component weights and highly accurate nutrition labeling,” they said in the paper.

“We demonstrate the potential of this data by training a computer vision algorithm that can predict the caloric and macronutrient values ​​of complex real foods better than professional nutritionists.”

SnapCalories technology is made possible by Roboflow, which has developed advanced “computer vision” systems. In Norris’ words, “Roboflow’s data management tools far exceeded any other tool we evaluated in the computer vision space.”

SnapCalorie uses Roboflow technology to manage, annotate, annotate datasets, manage tag-only users outside of its organization, and train computer vision models to annotate using models. Roboflow claims its technology helps SnapCalorie speed up model prototyping by 1,200% and reduce markup time by 80%.

Norris participated in Y Combinator and was a technical lead on the Google AI team that pioneered Google Lens and the Cloud Vision API. He is also the founder of Obico, formerly known as The Spaghetti Detective. The goal of the startup was to build smarter 3D printers using artificial intelligence and computer vision.

Press release