Count calories with a photo
Snap a photo of your meal and Fuse estimates what is on it: every item separately, with grams, calories, protein, carbs and fat. If a portion is not quite right, you adjust the grams. That way you log a meal in seconds, without looking everything up.
How it works
- Step 1
Take a photo
Of your plate, or pick one from your gallery.
- Step 2
Check the estimate
The AI lists each item separately, with grams, calories and macros. Adjust the grams, or remove what is wrong.
- Step 3
Add it all
One tap and your meal is in your day.
A photo, a barcode or a quick search.
Pick whatever is quickest for each meal. Everything counts toward today's calories and macros.
- Snap a photo, the AI recognizes your meal
- Scan the barcode on a package
- Over 1,700 Dutch foods, even offline
- Your calorie goal adjusts every week

How accurate is counting calories from a photo?
A photo is an estimate. An AI usually recognizes what is on your plate, but how much it is stays tricky: a photo has no depth, and oil, butter or sauce cooked into your food is not visible.
In recent studies, estimates from a photo alone were off by about a quarter to a third of the real calories per meal on average, and by more for some AI models. They were more often too low than too high, especially for large portions. When the AI also got the ingredients and amounts, it came much closer: in one study the error dropped from about 30 to 14 percent.
That is why Fuse shows each item separately with its grams, so you can quickly see whether a portion is right. If you know the amount, for example because you weighed it, enter the grams.
Logging by hand is not exact either: over whole days, people keeping a food diary in studies recorded 11 to 41 percent too few calories.
Photo, barcode or search: which one to pick?
For a plate of food you made yourself, a photo with a quick check is often the fastest. For packaged food, the barcode is usually more accurate.
| What you eat | What works best |
|---|---|
| Something from a package | The barcode. Fuse then uses that product's nutrition values per 100 grams; you enter how much you eat. |
| A plate you made yourself, or from a restaurant | A photo, then check the grams. |
| Something you weighed, or a single food like an apple or rice | Search the list of over 1,700 Dutch foods, even offline. |
| A package Fuse does not know | Your own food. Take a photo of the label, and the AI fills in the nutrition values printed on it. |
Tips for a good photo
- Get everything in the shot, with nothing hidden under something else.
- Shoot from slightly above, with your whole plate in the frame.
- Make sure there is enough light.
- Spread items out a little instead of piling them up.
- Cooked in oil or butter, or sauce or dressing on top? Add it separately with search afterwards: a photo usually does not show it.
- Check the grams, especially for large portions: those are more often underestimated.
Everything in one place
Fuse is more than a calorie counter. You also log your workouts and your runs in it, and you take on your friends every week. So one app shows what you eat and what you do.
Frequently asked questions
How does counting calories with a photo work?
Where does my photo go?
What if the AI gets something wrong?
Can I change it later?
Can I count calories without a photo?
Sources
- Coburn, B. et al. (2025). Comprehensive Evaluation of Large Multimodal Models for Nutrition Analysis: A New Benchmark Enriched with Contextual Metadata. arXiv 2507.07048 (preprint).
- Fridolfsson, J. et al. (2025). Performance Evaluation of 3 Large Language Models for Nutritional Content Estimation from Food Images. Curr Dev Nutr 9(10), 107556.
- Rodríguez-Jiménez, M. et al. (2025). Image-Based Dietary Energy and Macronutrients Estimation with ChatGPT-5: Cross-Source Evaluation Across Escalating Context Scenarios. Nutrients 17(22), 3613.
- O'Hara, C. et al. (2025). An Evaluation of ChatGPT for Nutrient Content Estimation from Meal Photographs. Nutrients 17(4), 607.
- Burrows, T.L. et al. (2019). Validity of Dietary Assessment Methods When Compared to the Method of Doubly Labeled Water: A Systematic Review in Adults. Front Endocrinol 10, 850.
