If you’ve used DailyBurn’s new app, MealSnap, you’ve probably been mystified wondering how it works. MealSnap is an app that lets its user take a picture of the food they’re eating and, based on that image, will provide calorie information for that food.
I’ve used app for about 6-7 meals now and I’ve been pretty amazed at the accuracy. So much so that I had to find out how they did it.
There has been some speculation at how it works in articles, but I couldn’t find a solid answer. Is it image recognition? Do they have a team? Do they use Mechanical Turk? Well… one of those hypotheses is incredibly easy to test.
I visited Mechanical Turk, typed in “food”, and ta-da!
Well.. truth be told, I had to hit refresh because these HIT’s are moving fast and the first search turned up no results. Looks like they’re using Mechanical Turk to identify the foods at anywhere from $0.02-$0.05 per picture and then using the data returned from Mechanical Turk to search for calorie information in their already well established database of food.
With a $2.99 price point for the app, DailyBurn would start losing money at around meal 60. By that point, however, Daily Burn has a loyal user that can easily be converted to the sale of another app in their family of products.
It’s actually refreshing to see such a non-technical solution after apps like IntoNow, Shazaam, Google Goggles and Word Lens have amazed me with new breakthroughs in analyzing the real world. Sometimes the most simple solution gets the job done. Maybe I should use manual labor in Self Compete?