FREE TOOLS
Protein estimator: what’s in this meal?
Estimates are approximate; adjust portions for accuracy. Education, not medical advice.
Updated August 25, 2026
IN SUMMARY
A protein estimator turns a meal into a number you can act on. Photograph the plate, describe it in words, or do both: an AI reads back the protein in grams, the range around it, and the items behind it, each one editable. Put any standard-sized loyalty or credit card face down (nothing personal or private showing) or a US quarter beside the plate, and your camera's photo is measured against it, which narrows the range; without a reference the portion is sized by eye and the range honestly widens. One meal at a time, free, no signup. And the photo is never stored: it is downscaled in your browser, sent once, and discarded.
How the protein estimator works
The AI identifies what is on the plate and how much of it there is; nutrient values come from USDA FoodData Central, the public reference database. The number that matters is the pair: a point estimate and its range. A tight range means the meal was sized with a reference in the shot; a wide one means the portion was judged by eye. Shoot from directly above for best sizing. And the card trick is privacy-safe by construction: the reference is painted out of the image on your device before upload, so what you sized the meal with never leaves your phone.
Two habits make any photo estimate better. Keep the protein visible: a camera cannot count chicken buried under a salad. And when the protein is buried by design (a stew, a curry, anything sauced), say so in the text box: one word about what's inside beats any photo angle. The items come back as a list with grams you can correct: treat the estimate as a starting point and your own portion knowledge as the editor. Corrections are recorded against the scan, the record we will use to improve the estimator over time. Estimates are approximate by nature; the honest use is trend, not truth, which is the same reason your daily target is computed from measured fat-free mass rather than guessed.
Hands full? Say it
The words box takes dictation: tap the mic, describe the plate while you carry it, and the transcript lands in the field for you to edit before sending. The estimate can be read back aloud too, and that voice runs on your device, not in a cloud: what you ate is a health detail, and speaking it should not require posting it to a vendor. Photo, words, or voice: the fastest input wins, and they stack.
Protein only, on purpose
The readout is protein in grams, not a calorie audit, not a judgment about the rest of the plate. On appetite-shrinking medication the scarce resource is stomach room, and protein is the macronutrient that decides whether the weight leaving is fat or muscle. One number per meal, sized against a daily target, is a habit that survives; a forty-field food diary is one that doesn't.
Where it fits: on a GLP-1, appetite shrinks before habits do, and protein is the intake worth defending; the keep-the-muscle playbook explains why. The estimator answers the meal-sized question; the app logs the day against a target and charts what the months did, on a readout you can trust.
Protein from a photo, answered
How accurate is a protein estimate from a photo?
Honest answer: it is an estimate, and the range around the number is the real reading. A clear overhead photo with a card or quarter beside the plate for sizing narrows the range; a photo sized by eye widens it. Portions you know beat any camera. The items come back editable so you can correct grams before you trust the total.
Do I need a photo, or can I just type what I ate?
Either works, and both together work best. A sentence like 'grilled chicken, about a palm, big bowl of rice' gets a real estimate on its own, and words beat the camera whenever protein is buried inside a stew, a curry, or anything sauced.
What happens to my photo?
It is downscaled in your browser before upload, sent once for the estimate, and discarded, never stored. If a reference card is in the shot, it is painted out of the image on your device before anything is uploaded.
Where do the protein numbers come from?
The AI identifies the items and sizes the portions; nutrient values come from USDA FoodData Central, the public reference database. Values are label- and preparation-dependent, which is one more reason the range matters more than the point number.
Does the app do this every day?
This page estimates one meal at a time and tracks nothing. Bild Health, the app, logs protein against a daily target sized to your measured fat-free mass. The estimator is the camera end of the same idea.
Know the number before you eat it.
Bild Health keeps fat mass, lean mass, protein, training and your dose history in one record — the lean number measured from your scale, anchored by imported DEXA scans.
Lose the fat. Keep the muscle.