AI food scanner: how a photo becomes calories, protein, carbs, and fat
How an AI food scanner reads a meal photo and estimates calories and macros — what computer vision gets right, where your description matters, and how accurate the numbers really are.
Logging every meal is the single hardest habit in fitness. Axiom's AI food scanner removes the friction: snap a photo, add a quick note, and get a full macro breakdown in under 10 seconds.
How the scan works
- Step 1 — Photo: your phone camera captures the plated meal.
- Step 2 — Vision model: an AI identifies each food item and estimates the visible serving.
- Step 3 — Refinement: you add anything the camera can't see (butter, oil, sauces, drinks, dessert, or the name of an unfamiliar dish).
- Step 4 — Personalized fit: the model re-scores the meal against your daily calorie and protein targets.
Why the second step matters
A vision model can see a plate of tacos, but it can't see two tablespoons of olive oil in the pan or the large soda you had on the side. Axiom always asks 'is there anything else we should know?' before finalizing. That's where accuracy lives.
Traveling or eating something unfamiliar?
On vacation and staring at a dish you've never heard of? You can type the name of the meal — or even just describe what you think you're eating — and Axiom will use it as extra context. The AI blends what it sees with what you tell it and gives you a best-estimate breakdown, plus a confidence level so you know how much to trust it.
Limitations, plainly
AI estimates are estimates. Restaurant food varies. Home cooks vary. Two beef tacos in California are not two beef tacos in Puebla. Axiom shows a confidence indicator on every scan and lets you edit any macro before saving. Treat the number as a good working estimate, not a laboratory reading.