There were also a few things it couldn’t recognize, but the suggestions were generally solid.

It had no trouble identifying a Jacobs Cream Cracker, for example, but homemade meals will inevitably be more of a challenge.

“It’s a deep learning algorithm, so we are using our user’s images to train it to be more accurate, and so it’s improving every day.”

For example, it recognized paella, but based the calorie estimate on a classic recipe, which resulted in a discrepancy of a few hundred calories, because our recipe uses less oil and slightly different ingredients. We found that Foodvisor correctly identified everything from bread to paella, but there’s still some work involved in accurate recording. and compared the results with our manual logging in MyFitnessPal. launch planned before the end of the year. It’s now expanding to the rest of the Europe starting with the U.K., Germany, Spain, and Italy, with a U.S. The app launched in France in 2018 and already has 1 million users. If you scan something that isn’t there already, you can add it yourself and it will recognize the item the next time you scan it. You can also scan barcodes, and Foodvisor currently has more than 700,000 barcodes in its database, with more added daily. “When we started, the app could only recognize five or six items, but now it can recognize more than 1,000 different food items,” Tran said. “Now it can recognize more than 1,000 different food items.”

Take a picture to translate