The calorie tracker for restaurant food that doesn't rely on a database.
Restaurant food breaks traditional calorie trackers — databases are wrong, portions vary, and barcodes don't exist. Nommie is a calorie tracker for restaurant food that reads your plate directly from a photo.
Restaurant meals are where most calorie trackers quietly fail. The database entry for 'chicken pasta' at a chain is usually missing, wrong, or off by a hundred grams. Portions vary by location. There are no barcodes. Nommie was built specifically to work on this kind of plate — the AI looks at the food you're actually eating and estimates calories and macros from the image, not from a lookup.
What makes Nommie different
Trained on real restaurant plates
The vision model is trained on plated meals — pasta, tacos, salads, bowls, burgers — not just packaged food labels. Restaurant dishes are the default case, not the edge case.
Portion estimation from the image
Restaurant portions are wildly inconsistent. Nommie estimates the actual portion on your plate from visual cues, so a double-size bowl doesn't get logged as a standard serving.
No barcode, no problem
Most restaurant food has no barcode, no nutrition label, and no reliable database entry. Photo logging bypasses all three.
Handles mixed dishes
Tacos with rice and beans, salads with five toppings, noodle bowls with sauce — the AI identifies the components together and logs the dish as one meal.
The typical restaurant experience with a legacy calorie tracker looks like this: you search for the dish, the database doesn't have the specific location's menu, you pick a 'similar' entry from a chain that isn't the one you're eating at, you accept a default portion size that may or may not match your plate, and you move on hoping the number is within fifty percent. Multiply that across a week of eating out and your tracking is closer to fiction than measurement.
A photo-based approach flips the problem. The AI doesn't need a menu or a chain match — it's looking at the actual plate in front of you. A bowl of pasta with red sauce at a local place logs the same way as one at a national chain, because the model is reading the food, not the restaurant. For anyone who eats out more than a few times a week, this is the difference between tracking that works and tracking that doesn't.
Common scenarios where this matters: pasta dishes where portion size varies dramatically between restaurants; tacos and burritos where the fillings change per order; bowls and salads with mix-and-match ingredients; shared plates and family-style meals; and anything off a specials menu that won't appear in a database for weeks. A photo handles all of these in one shot.
Frequently asked questions
Why do regular calorie trackers fail on restaurant food?+
Traditional trackers rely on database entries, which are often missing, outdated, or tied to a specific chain location that doesn't match where you're eating. Portion sizes are inconsistent, and most restaurant meals lack barcodes or nutrition labels entirely. You end up guessing, and the guess compounds across meals.
Does Nommie work for small, local restaurants?+
Yes — this is where photo-based tracking has the biggest advantage. The AI reads the plate, not the restaurant name, so a local spot's pasta logs the same way as a chain's. You don't need a menu entry or nutrition panel, which is how most independent restaurants operate.
How does it estimate portion size in a restaurant setting?+
The model uses visual references — plate size, utensils, drink glasses — to estimate portion volume. Restaurant portions are notoriously inconsistent, so estimating from the actual plate matters more than assuming a standard serving size that may be half or double what you were served.
What about shared plates or family-style meals?+
Take the photo of what ends up on your plate, not the shared dish. For tapas or family-style meals, log each serving as you take it. Nommie treats each photo as one meal, so multiple photos across a long dinner is the cleanest approach.
Try Nommie for free
Snap a photo, get instant nutrition — no manual logging.