How to track calories without weighing your food

You do not need a scale to track calories usefully. You do need to know which foods it is expensive to guess at.

By the Wellix team Last updated 9 July 2026 7 min read

Short answer

To track calories without a food scale, estimate portions using your hand as a reference (a palm of protein, a cupped hand of carbohydrate, a thumb of fat), photograph your meal and let an app estimate it, or weigh only the four or five calorie-dense foods you eat most often. Any of these will get you within a range that supports weight change, because consistency matters more than precision.

The cost is real but smaller than most people fear. A vision model estimating meals from photos alone was off by about 30% on energy, falling to about 14% once ingredients were named¹ — while unaided self-report in a classic study underestimated intake by 47%.² Not weighing is not the same as not knowing.

Why weighing feels mandatory, and is not

A food scale is the most accurate tool available to someone tracking intake at home, and nobody disputes that. The problem is that weighing has a completion rate. It requires you to be in your own kitchen, before you eat, with the packaging still to hand — three conditions that fail at every restaurant, every dinner at a friend's house, and every lunch eaten at a desk.

A method that is 10% accurate on the 40% of meals you manage to log is worse than a method that is 25% accurate on all of them. That is the trade the rest of this guide is about.

Method 1 — Hand portions

Your hand scales with your body, which makes it a surprisingly serviceable measuring device. The conventional reference points:

Reference	Approximates	Typical amount
Your palm (thickness and width, excluding fingers)	A protein portion	~85–115 g cooked meat or fish
Your cupped hand	A carbohydrate portion	~30–40 g dry / ~120–150 g cooked rice or pasta
Your closed fist	A vegetable portion	~100–150 g
Your thumb (tip to base)	A fat portion	~10–15 g — one tablespoon of oil, a small handful of nuts

The thumb is the one that matters. Protein and vegetable portions are forgiving: misjudging a chicken breast by 30 g costs you about 50 kcal. Misjudging oil by one tablespoon costs you about 120 kcal, and it is far easier to do.

Method 2 — Weigh four foods, estimate everything else

Calorie density is distributed extremely unevenly across the things people eat. A 20% error on a bowl of spinach is a rounding error. A 20% error on a pour of olive oil is a meaningful fraction of a day's deficit.

If you are willing to weigh a handful of foods and nothing else, weigh these:

Cooking oils and butter — roughly 120 kcal per tablespoon, invisible after cooking, and routinely poured rather than measured. This single habit recovers most of the available accuracy.
Nuts, nut butters and seeds — extremely dense, and portioned by handful.
Cheese — dense, and sliced by eye.
Dry grains, pasta and rice — cheap to weigh before cooking, and they triple in mass afterwards, which destroys visual estimates.

Everything else — vegetables, fruit, lean protein, most whole foods — can be estimated by hand or by photo with little consequence.

Method 3 — Photograph the meal

A vision model can identify the foods on a plate and estimate portions from the image. The published accuracy of this approach is around 30% mean absolute percentage error on energy when the model has only the photograph, improving to about 14% when it is also told what the ingredients are.¹

That second figure is the important one, and it tells you how to use any photo-based app: the app's guess is a starting point, and your correction is where the accuracy comes from. The camera can see surface area and identity. It cannot see the oil the vegetables were roasted in, whether the yoghurt was full-fat, or how much rice is hidden under the curry. You know all three.

The single highest-value correction. If you tell a photo app one thing about your meal, tell it the cooking fat. "Roasted in a tablespoon of olive oil" is worth more than any other sentence you could add.

Method 4 — Use packaging when it exists

For anything that arrives in a package, the label is strictly better information than a scale and better than a photograph, because it reports the manufacturer's own measurement. Read the serving size rather than assuming the package is one serving — this is the most common single source of large tracking errors, and it is entirely avoidable.

Method 5 — Anchor on the weight trend, not the daily number

Whichever method you use, your daily calorie total carries estimation error. Your bodyweight carries error too, mostly from water and gut contents. Neither individual number means much. The two-week trend of both, read together, means almost everything.

The practical loop is this:

Track your intake with a consistent method for two weeks. Consistency of method matters more than which method.
Track your weight over the same two weeks and take the average, not the last reading.
Compare the change in the average against what your calorie target predicted.
Adjust your intake target by 100–200 kcal in the direction the trend indicates, and repeat.

After two or three cycles you are no longer relying on a formula's estimate of your metabolism. You are relying on a measurement of it. That is the point at which the accuracy of any individual meal estimate stops mattering very much — and it is reachable without ever owning a scale.

One caveat about estimation error. The methods above have random error, which averages out over weeks. Unaided memory has systematic error, which does not: subjects in a New England Journal of Medicine study underreported their intake by 47 ± 16% and overreported their exercise by 51 ± 75%.² That study looked at obese, diet-resistant subjects specifically, so do not read the exact figure as universal. Read the direction as universal, because it is. Estimating a portion is fine. Reconstructing yesterday's meals from memory is not.

Where Wellix fits

Wellix is built for Methods 3 and 5. You photograph a meal, Wellix identifies each item and estimates calories, protein, carbohydrate and fat per item, and it shows you the itemised list before saving so you can add the oil or correct the portion. When an item is genuinely ambiguous, it asks you one clarifying question rather than guessing. Your weight trend sits alongside your intake in the same app, which is what makes the adjustment loop above practical rather than theoretical.

If you want to know exactly how much to trust a photo estimate, we lay out the published error figures — including where photo estimation fails badly — in how accurate are AI calorie counting apps.

Frequently asked questions

Can you count calories accurately without a food scale?

You can count them accurately enough to lose or gain weight reliably, which is a different and lower bar than measuring them precisely. Hand-portion estimation, photo-based estimation, and weighing only calorie-dense foods all produce errors that largely average out across a week, provided you apply the same method consistently and adjust based on your actual weight trend.

Which foods should I weigh if I only weigh a few?

Weigh cooking oils and butter, nuts and nut butters, cheese, and dry grains, pasta and rice. These are calorie-dense, routinely portioned by eye, and responsible for most large tracking errors. Vegetables, fruit and lean protein can be estimated by hand or photo at little cost.

How accurate is estimating portions with your hand?

Hand portions are typically accurate to within roughly 20–30% for a given food, which is adequate for protein, vegetables and most carbohydrates. They are least reliable for fats and oils, where a single thumb-sized misjudgement is worth about 120 kcal, so those are worth measuring properly.

Is a photo calorie app better than guessing portions?

Published evaluations put photo-only energy estimation at about 30% mean absolute percentage error, improving to about 14% when the user names the ingredients. That is comparable to careful hand-portion estimation and considerably better than reconstructing meals from memory, which carries a large systematic underestimation bias.

Do I need to track calories every day?

No. What you need is a consistent method applied often enough that your two-week average intake is meaningful. Tracking every meal with a rough method beats tracking some meals with a precise one, because the meals people skip logging are systematically the larger ones.

References

Rodríguez-Jiménez M, Martín-del-Campo-Becerra GD, Sumalla-Cano S, Crespo-Álvarez J, Elio I. Image-Based Dietary Energy and Macronutrients Estimation with ChatGPT-5: Cross-Source Evaluation Across Escalating Context Scenarios. Nutrients. 2025;17(22):3613. doi:10.3390/nu17223613. pmc.ncbi.nlm.nih.gov/articles/PMC12655113
Lichtman SW, Pisarska K, Berman ER, et al. Discrepancy between self-reported and actual caloric intake and exercise in obese subjects. N Engl J Med. 1992;327(27):1893–1898. pubmed.ncbi.nlm.nih.gov/1454084

Wellix provides general nutrition information and is not medical advice. It does not diagnose, treat, or prevent any condition. Calorie and macronutrient targets are estimates produced by predictive equations, and individual requirements vary. Consult a qualified healthcare professional before making significant dietary changes, particularly if you are pregnant or breastfeeding, or have a history of disordered eating, diabetes, kidney disease, thyroid disease, or any other medical condition.

Photograph it instead

Wellix estimates every item on the plate, then lets you correct the one thing the camera could not see.