The new app can determine the caloric content of a dish using a neural network from a photo
One of the creators of Google Lens, Wade Norris, and an aerospace engineer, Scott Baron, have launched a service called SnapCalorie. The app uses neural networks to calculate calories and display macronutrients in food based on a photo.
Norris notes that SnapCalorie is not the first of its kind, but the app can determine the calorie count in a dish with an accuracy of less than 20%. The service utilizes built-in depth sensors in smartphones' cameras to accurately measure portion sizes.
According to Norris, the developers have created an algorithm that allegedly surpasses humans in assessing the calorie content of food. SnapCalorie identifies the type of products in the photo and measures the size of each portion. The algorithm was trained on 5,000 images of different food items.
"We took care to expose it to diverse and challenging conditions that can be encountered in the real world, and we weighed each ingredient on scales," Norris said. He noted that SnapCalorie is likely most effective with American food since it was developed in the United States.
"We've found that people's interest in understanding what they consume is growing. The negative impact of processed foods is becoming increasingly evident every day. We've heard that our users really like SnapCalorie, especially when eating out, as many restaurants do not disclose the nutrition information of their dishes," he explained.