Hacker News new | ask | show | jobs
by jashmatthews 435 days ago
While technologically cool the app is bullshit. 90% accurate isn't accurate enough for the job and it's effectively making shit up.

Are you eating a 10% calorie deficit or a 10% calorie surplus? Cal AI can't tell you.

Not possible to know accurately enough from a picture. Potentially ever.
1 comments
bberenberg 435 days ago
My understanding is that even nutrition packaging is often off by up to 30%. Sorry, no source, just what I’ve heard a number of times. If I want to diet I typically build this in and it works well. If they’re actually off by 10% that’s not terrible.

But I tried one of these apps years ago and it went a step further than photos. It used the front facing camera on iPhones to build a 3d model of the food and measure its volume as well. Even that was off by more up to 50% not 10%.

The interesting thing I found, and it’s obvious when you read it but not when you’re trying to diet, is if you don’t layer food on top of itself or other food, you (and a camera based calorie counter) will have a much better understanding of how much you’re eating. Bowls / mounds of food will deceive you.

link
tekla 435 days ago
There is no way its off that much.

Standard for cutting is about 500 calories deficit, for 1lbs lost a week. Lets say 2500 calories daily standard. That's 20%. If food packaging was off by 30%, food nutrition planning would be worthless, but we know it isn't because we see fairly consistent results from weightlifters (assuming they're actually weighing their food and not eyeballing/using a PoS app like this)

link
yorwba 435 days ago
NIST says "in general NIST’s measurements are accurate to within 2% to 5% for nutrient elements (such as sodium, calcium and potassium), macronutrients (fats, proteins and carbohydrates), amino acids and fatty acids. Its measurements are accurate to within 5% to 10% for water-soluble vitamins (such as vitamin C) and 10% to 15% for fat-soluble vitamins (such as vitamin D)." https://www.nist.gov/how-do-you-measure-it/how-do-you-know-y...

Maybe there's something where they're off by 30%, but how many people even track how much vitamin D they get from food?

link