I have a metabolic cart, and the answer surprises me sometimes on the former - just standing at my desk all day I burn >3300kcal. I actually went to talk to the guy in the Casey Neistat video about food labels, and almost bought a bomb calorimeter. Calorie counting, while not useless, is not very helpful.
I know approximately how many calories I burn. In addition to using an Apple Watch I track my workouts.
And honestly this is good enough. If I target a 500 calorie per day deficit, I miraculously lose 1 pound per week. Which makes sense because simple arithmetic says that there are 3,500 calories per pound and 7 days times 500 equals 3,500.
It really is that simple. What on earth do you have a metabolic cart for? You wear that on your face all day every day tracking how much carbon dioxide you breathe out? And you really believe you’re burning 3,300 calories just standing? I could maybe believe that if you are in the morbidly obese BMI range.
Calorie counting is extremely helpful. Ask anyone who did it for the first time and suddenly realized they were eating 9 servings of potato chips at a time without even thinking about it.
If you can target enough of a deficit and not be hungry, good for you. It's hard for most people. The input data is wrong, the output data is wrong/unknown, and one can only be hungry for so long.
The surprising part was that it took the science guy 10 hours to calculate the calories for 5 food items ? Would be nice if we could get that down to minutes and accessible to personal consumers.
Bomb Calorimetery is a PITA, I almost did a startup around this. The researcher in the video has a fairly manual one, there are automated setups that reduce this time, but they still cost $40k+ and require oxygen tanks, etc.
The even more difficult case for calories is something like a steak: how marbled is it? I could see some computer vision approach working.