Gayan Perera

Calorie tracking apps that calculate calories and macros from meal photos are becoming very popular. But they are not accurate yet.

Even under very controlled conditions — standardized plates, lighting, and camera angles — the error rates are large (Fridolfsson et al., 2025):

• Calories: ~36–64% error
• Food weight: ~36–65% error
• Carbohydrates: ~48–73% error
• Protein: ~61–110% error
• Fat: ~42–90% error

Accuracy is worse with larger portion sizes.

To put that into perspective: if a meal actually contains 700 kcal, a 36–64% error means the app could estimate anywhere from roughly 450 to 1,150 kcal.

AI can often recognize foods reasonably well. The biggest limitation is portion size estimation. Hidden ingredients make this even harder. For example, visually it’s almost impossible to tell whether a meal contains one tablespoon of oil or several. This is one of the same reasons people often underestimate calories when tracking food manually.

Accuracy is generally better when foods are simple and separated (Cofré, S., et al., 2025).

If this technology becomes more accurate (and hopefully it will), it would be great. It could eliminate the inconvenience and time consumption of manual tracking — which is one of the main reasons many people don’t want to track or give up after a few days. Taking a photo is far easier than entering food manually and could help more people stay consistent with tracking.

But the technology isn’t there yet. For now, it’s best to treat photo-based calorie estimates as rough guidance, not precise nutrition data.

References
Fridolfsson, J., Sjöberg, E., Thiwång, M. and Pettersson, S. (2025) ‘Performance evaluation of 3 large language models for nutritional content estimation from food images’, Current Developments in Nutrition, 9(10), p. 107556.

Comment “BOOK”👇 if you’d like a free copy, and I’ll send you the download link directly.

Over the years, many people have asked me to recommend a good nutrition book — something that’s accurate, practical, and easy to apply to how we actually eat.

I could never find one that I could confidently recommend.

Some nutrition books are good, but they’re written from a very theoretical angle and completely lack practical application. Others go to the opposite extreme — oversimplifying everything with “no carbs,” “no sugar,” or “cut out entire food groups” type of advice.

So, I decided to fill that gap — to write a book that’s scientifically accurate, easy to understand, and practical enough to apply to real Sri Lankan food and everyday life.

The Sri Lankan Fat Loss Guide is the closest thing to a complete, evidence-based framework for Sri Lankans who want to lose fat and keep it off — without giving up rice and curries.

This book was written using 122 peer-reviewed studies and reviewed for scientific accuracy — reflecting the most up-to-date evidence available, as close to 100% accurate as current research allows.

For context, among the 100 best-selling nutrition books, only about one-third even cited scientific research to support their claims, and most were riddled with inaccuracies or lacked evidence entirely (Loung et al., 2022).

Here’s what you’ll find inside:
🍚 How to calculate your calorie and macronutrient requirements
🥥 How to include foods like rice, coconut milk, and curries while still losing fat
⚖️ How to adjust your nutrition when progress slows
💪 How hormones like insulin, cortisol, thyroid, and testosterone actually fit into fat loss
🍷 How to handle social events, alcohol, and travel without losing progress
🧠 How to manage cravings, avoid burnout, and maintain results for life

This book explains the entire fat loss process — from beginning to end — in a simple, practical, and evidence-based way, built on 13 years of coaching over 1,300 Sri Lankans across 15 countries.

If you’d like a free copy, comment “BOOK” below 👇 and I’ll send you the download link directly.