MyNetDiary vs CalEye: Which Tracker Wins for Accuracy?

MyNetDiary has been a respected name in calorie-tracking software since 2008. It offers a large verified food database, barcode scanning, macro tracking, and dedicated diabetes tools built around carbohydrate estimation and glucose log integration. CalEye launched with a different hypothesis: that the most important accuracy problem in food logging is not database size, but the gap between what’s in the database and what’s actually on the plate. These are different bets about where logging fails most often. This comparison examines each claim against actual behaviour, publicly available feature specifications, and the body of research on dietary self-monitoring accuracy — the field that makes clear why database breadth alone is an insufficient measure of a tracking app’s real-world usefulness.

Database Size: What the Numbers Actually Mean

MyNetDiary’s flagship claim is a verified food database of over 1.4 million items, including branded products, restaurant chains, and international foods. The size is genuine and relevant for one specific logging scenario: foods with barcodes or restaurant menu items from major chains that have committed to nutritional disclosure. In those cases, scanning the barcode or searching by restaurant and menu item retrieves a verified calorie and macro figure that is accurate within the tolerance of the data the brand submitted.

The limitation is the scope of that scenario in real eating behaviour. The PREDICT study from Spector and colleagues at King’s College London, which tracked real-world eating behaviour across thousands of participants, found that the majority of meals consumed do not come from packages with barcodes or from chain restaurants with published nutritional data.¹ Home-cooked meals, restaurant dishes at independent restaurants, mixed dishes, and foods eaten at other people’s homes constitute a large fraction of actual dietary intake for most populations. For these foods, a database of 1.4 million items is still dependent on the user accurately identifying a database entry that matches what is on the plate — and that identification requires portion estimation, which is where accuracy degrades.

CalEye’s approach is to photograph the plate and let computer vision estimate the food items and portion weights directly. The database is smaller by item count but the identification mechanism is different: the image contains portion information that text-based database search does not capture. The trade-off is that image recognition is imperfect, particularly for unfamiliar or composite dishes, while database search is reliable when the right item exists and is correctly identified.

AI Logging and Photo Recognition

MyNetDiary introduced a photo recognition feature in later versions that uses image analysis to suggest food items for manual confirmation. The feature returns food suggestions from its database based on visual identification, which the user then selects and portion-adjusts manually. The recognition quality for common packaged foods and standard Western meals is reasonable, but the approach requires the user to manually confirm portions, which reintroduces the estimation problem.

CalEye’s photo recognition is built around portion weight estimation as the primary output, not just food category identification. The model estimates gram weight per food item from visual cues — plate diameter as scale reference, food geometry, depth estimation — and links each item to its USDA FoodData Central or SR-Legacy database entry. The output includes a stated confidence range: a logged item appears as “chicken breast (grilled), estimated 145 g ± 15 g, 175 kcal.” The explicit uncertainty disclosure distinguishes this from tools that present rounded integers without sourcing.

The USDA traceability is meaningful for users who want to audit their logs or compare to research nutrition data. When a CalEye estimate is linked to a specific USDA FoodID, a user or clinician can check the nutrient profile against the source rather than trusting a black-box database entry of uncertain provenance. This matters more for clinical users — dietitians reviewing patient logs, endocrinologists checking carbohydrate estimates — than for casual users, but it represents a different quality-of-data standard.

Diabetes-Specific Tools

MyNetDiary has invested significantly in diabetes tooling. Its Diabetes PRO tier (a paid subscription) offers direct glucose log import from connected glucometers and CGM devices including some Dexcom and Abbott FreeStyle Libre integrations, carbohydrate budget tracking against an ADA-aligned daily target, and A1c trend estimation from self-reported glucose values. A deeper look at these features — and how they compare directly to MyFitnessPal’s diabetes capabilities — is covered in MyNetDiary vs MyFitnessPal for diabetes. The integration between food log and glucose log is the most clinically useful feature: a user can see the postprandial glucose response plotted against the logged meal and identify which meal types correlate with glucose excursions.

The limitation of MyNetDiary’s diabetes tooling is the underlying accuracy of the carbohydrate estimates it feeds into glucose correlation analysis. If a user logs “pasta, restaurant, medium portion” and the database returns 65 g of carbohydrate, the correlation between that meal and the subsequent glucose reading is only meaningful if 65 g was an accurate estimate. Restaurant portion variance can be 30–50% in either direction, meaning the glucose correlation analysis is built on estimated data that may differ substantially from the actual ingested quantity.

CalEye’s diabetes-relevant features centre on glycaemic load display alongside carbohydrate grams — each food item is shown with its estimated GL based on the USDA carbohydrate figure and published glycaemic index data from the Sydney GI Database and Atkinson 2008 GI tables.² This gives carbohydrate-counting users immediate feedback on not just how many grams of carbohydrate are in a meal but how quickly that load will raise blood glucose. The ADA Standards of Care acknowledge GL as a useful supplementary metric for glycaemic management beyond gram counting alone.³

Barcode Scanning and Packaged Food Accuracy

For packaged foods, MyNetDiary is better. Its barcode database coverage for branded products in the US, UK, Canada, and Australia is extensive, and the scan-to-log workflow is fast. The accuracy for these items is as good as the manufacturer’s stated nutrition label — which is itself accurate to within the FDA-allowable tolerance of 20% for most nutrients. For daily calorie-budget tracking of packaged food diets, this is the most reliable category.

CalEye’s barcode scanning exists but the database is smaller than MyNetDiary’s for this category. The product’s comparative advantage is in unpackaged, cooked, and restaurant foods — exactly the category where MyNetDiary’s database can only offer a text-based search and manual portion estimation. Whether this matters to a given user depends entirely on the composition of their diet. A person who eats predominantly packaged and prepared foods from major brands will get better logging throughput from MyNetDiary’s barcode-first workflow. A person who eats predominantly home-cooked meals, restaurant food from independent restaurants, or traditional cultural cuisine will get better portion accuracy from photo-based logging.

Long-Term Adherence: Where Both Tools Face the Same Challenge

The most consistent finding in dietary self-monitoring research is that adherence to food logging declines sharply over time, regardless of the tool. Research on what makes a calorie tracking habit stick identifies the specific friction points — logging speed, meal variety, and out-of-home eating — that determine whether users are still logging at month three. Lichtman’s 1992 NEJM study established that even motivated, supervised participants systematically underreport intake.⁴ More recent research from the NWCR — tracking people who maintained large weight losses for more than a year — finds that continued self-monitoring is among the strongest predictors of weight maintenance, but that the monitoring method matters less than the act of monitoring itself.⁵

Both MyNetDiary and CalEye face the same structural challenge: logging friction determines long-term compliance. MyNetDiary’s database breadth reduces friction for barcode foods. CalEye’s photo capture reduces friction for unpackaged meals. Neither eliminates the motivation problem inherent in self-monitoring over months and years. The tools are complements, not competitors, in this sense — a user who combines photo logging for meals with barcode scanning for snacks and packaged items gets the friction reduction benefit of both approaches.

The user interface matters for adherence in ways that are difficult to quantify in a feature comparison. MyNetDiary’s interface is mature and comprehensive — it surfaces a large amount of information and has extensive customisation options, which experienced trackers find valuable and new users can find overwhelming. CalEye’s interface is newer and more streamlined, optimised around the photo capture workflow. The right choice depends on the user’s experience level and tracking style.

Pricing and Access

MyNetDiary’s free tier includes basic calorie tracking with database search and barcode scanning. The Premium tier adds macro targets, meal planning, and expanded reporting. The Diabetes PRO tier adds glucose integration and carbohydrate budget management. At the time of this comparison, MyNetDiary Premium runs approximately $6–8 per month on an annual plan.

CalEye’s pricing structure is available on the app’s product page. The relevant comparison point is not price alone but price relative to the accuracy improvement for the specific foods a given user eats most often. A $6/month tool that accurately tracks 70% of your meals is a better investment than a free tool that accurately tracks 40%.

The Accuracy Verdict

The question this comparison started with — which tracker wins for accuracy — does not have a single answer because accuracy is food-type dependent. For packaged and branded foods with barcodes, MyNetDiary is more accurate because it has broader database coverage and a reliable scan-to-log workflow. For unpackaged meals, restaurant food, home cooking, and mixed dishes, CalEye’s photo-based portion estimation is likely more accurate because it captures portion geometry information that text-based database search cannot.

The systematic underreporting problem documented in Lichtman 1992 and replicated by Urban 2011 is primarily a problem of unmeasured portions and omitted items — cooking oils, sauces, condiments — that don’t get logged because the logging workflow is too slow or requires too many steps. The guide to hidden calories in dressings, sauces, and oils quantifies exactly how large these omissions tend to be across common meal types.⁴ Photo logging reduces the omission problem significantly: a photo captures the plate as seen, including the visible oil on the stir-fry and the sauce on the protein. It doesn’t eliminate uncertainty about hidden ingredients (butter in restaurant cooking, sugar in sauces), but it makes the visible portion more accurate than a text search with manual volume estimation.

For most real-world users whose diets are a mix of packaged foods and cooked meals, the optimal strategy is to use the tool whose strengths match the majority of their logged meals, and to recognise that no tool eliminates the fundamental accuracy ceiling imposed by cooking variability and visual portion estimation. Both MyNetDiary and CalEye are more accurate than the alternative — not logging at all.

References

1. Spector T, Berry SE, Drew D, et al. “Human Postprandial Responses to Food and Potential for Precision Nutrition: The PREDICT Study.” Nature Medicine 26, no. 6 (2020): 964–973.

2. Atkinson FS, Foster-Powell K, Brand-Miller JC. “International Tables of Glycemic Index and Glycemic Load Values: 2008.” Diabetes Care 31, no. 12 (2008): 2281–2283.

3. American Diabetes Association Professional Practice Committee. “Facilitating Positive Health Behaviors and Well-being to Improve Health Outcomes: Standards of Care in Diabetes — 2024.” Diabetes Care 47, Supplement 1 (2024): S77–S110.

4. Lichtman SW, Pisarska K, Berman ER, et al. “Discrepancy Between Self-Reported and Actual Caloric Intake and Exercise in Obese Subjects.” New England Journal of Medicine 327, no. 27 (1992): 1893–1898.

5. Wing RR, Phelan S. “Long-Term Weight Loss Maintenance.” American Journal of Clinical Nutrition 82, Supplement 1 (2005): 222S–225S.

6. Urban LE, McCrory MA, Dallal GE, et al. “Accuracy of Stated Energy Contents of Restaurant Foods.” JAMA 306, no. 3 (2011): 287–293.