Tracking food · 3 minute read
A photo is a starting guess
A photo can help identify foods and estimate portions. It cannot see every ingredient or measure hidden oil. Norman lets you review and edit the estimate before saving it.
A simple example
Two similar-looking rice bowls can contain different amounts of rice, dressing and oil. A note like “half the rice, sauce on the side” can help, but it is still an estimate.
Your next step
Check the food names and portions. Add missing ingredients, remove items you did not eat and change the calorie estimate if you know the label.
A little more detail
Image angle, food hidden under other food, plate size and recipe variation all affect accuracy. A high confidence label is not a guarantee. Database entries vary with brands, preparation, serving sizes and data quality.
Questions you might have
How accurate is AI photo estimation?
It is useful for a first guess, not an exact count. Hidden oils and portion depth are hard to see. Review before saving and use labels or recipes when you have them.
Why do food databases disagree?
They may describe different brands, raw versus cooked food, serving sizes or recipes. Some entries are user-submitted. Match the description and amount to what you ate, and prefer the package label for that product.
What if Norman identifies my meal incorrectly?
Edit or remove the wrong item before saving. Add a short description or enter the meal manually. Never feel obliged to accept the photo’s guess.
Estimates and general education, not a medical prescription. Seek individual advice when your health or needs call for it.