SnapCalorie uses artificial intelligence to estimate the calorie content of food from a photo

Former Google employee Wade Norris and engineer Scott Baron have created SnapCalorie, an app that measures calories and macronutrients from a single photo.

Here's What We Know

According to the developers, SnapCalorie combines many new technologies and algorithms to count calories. It also uses depth sensors on supported devices to measure portion sizes and a team of human reviewers for "an added layer of quality".

The developers used a dataset of 5,000 meals to train the algorithms. It contains thousands of photos of each culinary item, such as soups, burritos, butter, "mystery sauces" and others.

The results can be logged in a food journal or exported to fitness tracking platforms.

Norris admitted that SnapCalorie's algorithm may be biased towards US food, as the team has collected most of the raw data for training in the US. But the company is in the process of expanding the dataset to include other cultural cuisines.

Norris doesn't claim that SnapCalorie is 100% accurate, suggesting that the app's calorie-counting tools are only part of a larger nutrition puzzle.

Source: TechCrunch.