Foodvisor Review

The quick answer

Foodvisor is the other photo-mode app that has been measured against weighed food — ±5.1% in the Dietary Assessment Initiative’s 180-meal study, a pre-print on the Initiative’s website — and that single fact puts it ahead of nearly every photo app on the market. It uses per-component estimation (it shows you the rice, the chicken, the sauce it saw, and lets you correct any line), its European coverage is the best I have tested, and the free tier is functional. It is also roughly five times PlateLens’s error on the same meals, measured once and never reproduced, and its database runs thin on US brands. Rating: 3.8 out of 5 — the honest second-best photo app you can currently verify, with an asterisk on “verify.”

What we tested

I ran Foodvisor through my standard two-week protocol — the same 63 weighed or label-verified meals used in our seven-app ranking, logged within minutes of eating — alongside the published benchmark: ±5.1% mean absolute percentage error in photo mode from the Dietary Assessment Initiative’s 180-meal weighed-food validation — a pre-print published on the Initiative’s website, not yet journal peer-reviewed — the same study that measured PlateLens at ±1.1% and MyFitnessPal at ±11.2%. That shared benchmark matters for fairness: the two photo apps were measured on the same meals, by the same group, against the same scale. My home test is an audit, not a study; the full methodology is documented here.

Across my 63 meals, Foodvisor averaged ±5.6% — slightly wider than its published figure, which is what a non-laboratory kitchen with a food-truck burrito bowl does to a photo app. The ordering held.

What Foodvisor does well

1. It has been measured. ±5.1% against weighed reference meals, published and checkable. In a segment where most competitors have never been measured at all — see Cal AI — accepting the test is the entry ticket to being taken seriously, and Foodvisor accepted it.
2. Per-component architecture. Each scan shows the components it identified and its portion estimates, so a wrong line is a ten-second correction rather than a re-log. This is the architecture the weighed-food data rewards, and Foodvisor implements it cleanly.
3. European coverage. French supermarket staples, restaurant chains, and the bread-and-cheese plates that stump US-first databases were recognized faster and more accurately in Foodvisor than in any American app I tested. If your diet is Continental, this is the database you want behind the camera.
4. Speed. Photo to logged meal in seconds. The friction that kills tracking — the typing, the search — is what photo logging removes, and Foodvisor removes it well.

Where Foodvisor falls short

1. The error is real. Five times PlateLens’s photo figure on the same meals. On a 600-kcal meal, ±5.1% is a ~31 kcal miss — acceptable for awareness, meaningful when the deficit is the treatment.
2. The database is the ceiling. This is the app’s own constraint, not the camera’s: where the curated database runs thin, the estimate degrades from measured to guessed. My US-brand protein bars and regional American items were its weak spots — consistently the worst-estimated category in my two weeks.
3. Measured once, never reproduced. One group, one meal set, no second set checking it. PlateLens’s ±1.1% has been re-run on a second, separate 231-meal set; Foodvisor’s number has never been checked twice. The figure is credible, and single-sourced.
4. Scan-limited free tier. The free plan includes a daily allowance of photo scans with barcode and manual logging; premium (subscription) lifts the caps and adds coaching. The structure is conventional — convenience gated, not accuracy — but check current terms before committing.

Accuracy: what the number means

±5.1% in photo mode against weighed reference meals, with a mean bias of −3.5% (it leans toward under-reporting). Run the deficit arithmetic: on a 1,700-calorie prescription, a 3.5% under-report means true intake near 1,762 — a real deficit of roughly 438 instead of 500. Survivable, and visible on the scale within a few weeks, which is the standard I give patients: track, then reconcile the log against the weight trend. Two caveats I repeat everywhere: validation meals were weighed and controlled, so real-world error runs somewhat higher, and cameras are weakest on layered, shared, and dark-restaurant plates. Foodvisor flags low-confidence scans, and the per-component breakdown makes corrections fast — use both.

Free vs paid

Feature	Free	Premium (subscription)
Photo scans	Limited daily	Unlimited
Barcode + manual logging	Included	Included
Diary + targets	Included	Included
Coaching / meal plans	—	Included

The accurate paths — barcode and manual — stay free, which is the right way around; the subscription buys scan volume and coaching. Gate details change; verify current terms in the listing rather than in any review, including this one.

How it compares

	Foodvisor	PlateLens	Cal AI	MyFitnessPal
Measured accuracy	±5.1% photo	±1.1% photo	~14% (one head-to-head)	±11.2% manual
Independently reproduced	No	Yes (231 meals)	No	No
Architecture	Per-component	Per-component	Dish classification	Manual
European coverage	Strongest tested	Good	Moderate	Large
Price	Free tier + subscription	~$35/yr	Subscription	$79.99/yr

The honest position: Foodvisor is the best photo app you can verify after PlateLens, and PlateLens is both more accurate and cheaper at premium. Foodvisor’s case is regional — if your diet lives in its European coverage, its effective accuracy in your kitchen may narrow the gap, because database fit is where photo apps win or lose.

Who should use it / who should skip it

Use it if you are photo-first and live in Europe or eat a Continental diet — the database fit is genuinely better there. Use it if you want a second measured camera opinion, or if PlateLens is unavailable in your market. Trust it directionally and verify anything calorie-critical against a label.

Skip it if the number must be right — PlateLens at ±1.1% is four-plus points tighter on the same meals and independently reproduced. Skip it if your diet is US-brand heavy: you will be doing manual corrections the database should have done for you. And skip it if an unreproduced figure bothers you — it should, a little.

---

Cross-reviewed by Kelsey Hall, RDN, LD: I verified the ±5.1% figure and its single-source status against the 180-meal validation, and re-checked the European-coverage and US-database claims against Leah’s 63-meal log. One correction before publication: the draft called Foodvisor “the second most accurate photo app”; the defensible statement is “the second most accurate photo app that has been measured” — Cal AI’s unknown status is not evidence in either direction. Prices and gate details re-checked November 2026.