Calorie scanner app: what it does, and why it needs no install

What a calorie scanner app actually does

Three things happen between pointing your phone at a packet and seeing a calorie figure, and they are worth separating because apps fail at different ones.

Decoding. The camera reads the striped symbol and turns it into a number — usually 13 digits in Europe, 12 in North America. This is pure optics and maths, it has been reliable since the 1970s, and no app is meaningfully better at it than another. If a scan fails here, the cause is light, focus or framing, not the software.

Lookup. That number is matched against a food database. This is where apps genuinely differ. A barcode is only an identifier; it carries no nutrition information at all. Everything you see afterwards depends entirely on somebody having recorded that product's panel against that number.

Portioning. The database holds figures per 100 g. You are about to eat a 45 g portion. Converting between the two, correctly, is where most of the real-world error creeps in — and where a lot of apps quietly hand the problem back to you.

The database is the product

Every calorie scanner is a front end onto a food database, and the quality of that database determines whether the app is useful in your particular supermarket.

There are three sources in common use. Manufacturer feeds are accurate but narrow, covering only brands that have bothered to supply data. Crowd-sourced databases like Open Food Facts are enormous — around three million products — and cover own-brand and regional items that no manufacturer feed will ever include, at the cost of occasional errors and gaps. User-submitted entries inside a closed app are the worst of both: unverifiable, frequently duplicated, and often wrong by a factor that nobody catches.

This site uses Open Food Facts, and says so. Coverage is strongest in Western Europe, good in North America, and thinner elsewhere. When a product is not there, we tell you it is not there rather than showing you a similarly-named item and letting you assume it matched.

Why a scanner does not need to be an installed app

The camera, the decoder and the database lookup are all available to a web page now. A modern browser can open a video stream, decode a barcode in real time on-device, and query a database — which means the entire "app" can be a URL.

The practical differences are all in your favour:

• Nothing to install. No 80 MB download to scan one tin of beans.
• Nothing to sign up for. A web page can simply work. An app store listing has a business reason to collect your email first.
• Nothing running in the background. Close the tab and it is gone. No background refresh, no notifications, no location access it never needed.
• Always current. There is no version of this page that is three updates behind.

The honest trade-off: a web scanner cannot send you a reminder, and on iOS it needs Safari to reach the camera. If your priority is a daily nudge at 8pm, an installed app still wins. If your priority is finding out how many calories are in the thing in your hand, the browser version answers faster.

What separates a good scanner from a bad one

It tells you when it does not know

The single most damaging behaviour in a calorie app is a confident wrong answer. A scanner that returns the nearest similarly-named product rather than admitting the barcode is unknown will quietly corrupt a whole week of logging. Ours returns nothing and offers manual entry, which is less satisfying and much more useful.

It defaults to the portion you are actually eating

Database figures are per 100 g. If a scanner shows you the per-100 g number and moves on, it has handed you an arithmetic problem at the exact moment you are least inclined to do one. A good scanner defaults to the manufacturer's stated serving, shows what that serving weighs, and lets you change it with the figures updating live.

It shows macros, not just calories

Two 400-calorie lunches can be completely different meals. Protein, carbohydrate and fat are the difference between "I ate 400 calories" and "I understand what I ate", and a scanner that hides them behind an upgrade prompt is withholding the part that matters.

It handles food without a barcode

Most of what people actually eat — a cooked meal, a restaurant plate, a piece of fruit — has no barcode at all. A scanner that only does packets covers a minority of your diet. That is what the AI calorie scanner is for.

Using it well

Scan the packet before you open it, not after. Once a wrapper is in the bin you are relying on memory for the brand and the size, and own-brand products vary more between shops than people expect.

Check the portion the scanner assumed. This is the one habit that removes most logging error. A "serving" of cereal on the box is frequently 30 g, which is a good deal less than what ends up in an actual bowl.

Do not chase precision you cannot have. Logging is useful because it shows you patterns across a week, not because any individual figure is exact. Spending four minutes deciding between 180 and 195 calories is time spent on noise.

You can try all of this now — the calorie scanner runs in this browser with nothing to install, and your first scan is free.