SnapCalorie, an innovative health-centric startup, has been making waves with its unique application of artificial intelligence (AI).
The brainchild of Wade Norris, a former Google employee and co-creator of Google Lens, and Scott Baron, an aerospace industry systems engineer, SnapCalorie uses AI to determine the caloric and macronutrient content of meals from a single smartphone photo.
This groundbreaking venture has recently raised $2 million in a funding round, which included notable investors like Accel, Index Ventures, and Y Combinator, adding to the $125,000 raised in a previous pre-seed round.
While SnapCalorie isn't the pioneer in using computer vision for calorie counting, it sets itself apart through its unique approach.
It leverages depth sensors on compatible devices to gauge portion sizes and employs human reviewers to ensure quality, thereby reducing the margin of error in calorie estimation to less than 20%.
This is an improvement over other AI-driven meal tracking apps that don't focus on portion size estimation, a critical factor in reducing errors.
Despite the promising technology, it's worth noting that AI-based calorie counters have faced skepticism due to issues with accuracy. For instance, a 2020 study found that even the best among these apps, Calorie Mama, was only correct 63% of the time.
SnapCalorie has taken steps to address these issues by developing a proprietary algorithm that performs better than a human in estimating a meal's caloric content. This is achieved by identifying the food items in a photo and calculating their portion sizes.
The algorithm's effectiveness is attributed to the unique training dataset of 5,000 meals, compiled through a process that involved taking thousands of photos of various meals with a robotic rig.
However, SnapCalorie acknowledges certain limitations. The algorithm shows a bias towards American food due to the primarily U.S.-based initial training data. But the company is working on diversifying the training data to encompass a broader range of cuisines.
Norris also stresses that no app can fully account for the many variables in caloric intake, such as differences in cooking methods. Therefore, SnapCalorie's calorie estimating tools are best seen as one piece of a larger nutrition management strategy. To further assist users, the app includes a chatbot, powered by ChatGPT, which offers meal suggestions based on a user's personal goals and past preferences.
SnapCalorie has found a particular niche among diners who frequent restaurants that don't provide nutritional information. Norris reports that the app is seeing a steady increase in popularity, with the company aiming to attract 1,000 new users this month alone. For now, SnapCalorie's focus is on expanding its user base, rather than monetizing its services, with Norris describing the company's spending as "very conservative".
