How Nutrola's AI calorie counting works

Last updated 28 जून 2026

When you scan or describe a meal, Nutrola sends it to Nutrola's servers, which identify the likely ingredients and return a calorie count plus protein, carbs, and fat. These are estimates, not exact measurements, so you can edit any value, fix ingredients, or describe what was off before saving. Barcode products instead pull their data from a large product database.

What happens when I scan a meal photo?

When you tap the center + in the tab bar and choose Scan Food, you take a photo with the live camera (tap the round shutter), or tap the gallery icon to import an existing picture. Nutrola uploads that image and analyzes it on its servers. It breaks the meal into individual ingredients and returns estimated calories along with protein, carbs, and fat for the whole meal. The result is a starting point you can adjust, not a final measurement.

How to log a meal Log a meal from a photo already in your gallery

How does describing a meal work?

Describe Food lets you say or type what you ate (for example, "two scrambled eggs with a slice of whole wheat toast"). Voice input is used by default, and you can toggle to type instead. Spoken descriptions are transcribed on Nutrola's servers, and then the same kind of analysis runs: Nutrola identifies the ingredients and estimates the calories and macros for you to review and edit.

Is the AI exact? How accurate is it?

No. Nutrola produces estimates. Real food varies in size, preparation, and hidden ingredients (oils, sauces, dressings), so the AI is making its best guess from what it can see or read. Treat the numbers as a close starting point and correct anything that looks off. Consistent logging over time matters more than any single perfect entry.

How do I fix a wrong estimate?

You have a few ways to correct an AI result before or after saving:

  • Tap the calorie value or any macro to edit it directly (Edit Calories, Edit Protein, Edit Carbs, Edit Fat).
  • Edit, add, or remove individual ingredients.
  • Tap Fix Result and describe what was wrong in plain language (for example, "I only ate half of the portion" or "it was made with almond milk"). Nutrola re-analyzes with that correction.

The calorie or macro estimate looks wrong Correct an AI meal with 'Fix Result' (tell it what's wrong) Edit a meal's calories, protein, carbs, or fat by hand Edit, add, or remove ingredients in a scanned meal

Why can't I set a portion size on a photo or voice meal?

AI photo and voice meals do not have a portion or serving-size control. They are stored on a fixed basis, so to change the amount you edit the calorie or macro values, edit the ingredients, or use Fix Result to say something like "I only ate half." Portion and unit switching (serving, g, oz, ml) is available only for barcode products.

Why can't I change the serving size of a photo-scanned meal?

How is barcode scanning different?

Scan Barcode does not use AI estimation. It looks the product up in a large product database and pulls that product's listed nutrition. Because the database does not include every product, some are not found. When a barcode is not found, Nutrola shows a card titled "Hmm, I don't know this one!" with a "Describe it instead" button so you can log it by description instead. A missing product is a coverage gap in the database, not a problem with your phone.

Barcode won't scan My photo isn't being recognized

Tips for better estimates

  • Get the whole plate in frame, with good lighting, before you take the photo.
  • Mention quantities and key ingredients when you describe a meal (sizes, sauces, oils, milk type).
  • Review the result and edit anything that looks high or low before saving.
  • For packaged items, scanning the barcode is usually more precise than a photo. You can also search the food database directly.

Log from the food database, your foods, and saved foods

Still need help? In the app, open Profile/Settings → Support & Legal → Contact us to chat with our support assistant, or email us at support@nutrola.app.

Was this article helpful?

Related articles

Still need help?

Our support team is here for you. Start a chat with our assistant or reach a human.

Contact support