Good morning, there 👋
No meals logged today.
Head to Track Food to get started.
Track Your Food
Scan food with AI camera, or search from 15,000+ verified foods
Take a photo or upload one
AI will identify the food
then click Scan with AI
Your History
Past 30 days of nutrition data
| Food | Meal | Calories | Fiber · Salt | Date |
|---|
Your Profile
Manage your goals and account
Export logs, calories, and 82+ micronutrients in Apple HealthKit format or import XML data.
Sync active calories burned, daily steps, and workout sessions directly into your daily energy balance.
Allows registered dietitians or physicians to lock individualized metabolic floors, set CKD protein restrictions, or customize clinical micronutrient targets.
🔬 Accuracy Benchmark
200-Meal International Reference Validation Suite v3.0
| # | Meal | Category | Ref Cal | Est Cal | Cal Err% | Pro Err% | Source | FDC ID |
|---|
Reference Standard: USDA FoodData Central SR Legacy, Indian Food Composition Tables (IFCT) 2024, NIN Hyderabad Food Composition Tables
Evaluation Type: Reference-database comparison against lab-calibrated nutritional profiles across 7 cuisine categories and 200 internationally-sourced meals.
Data Sources:
- USDA FoodData Central SR Legacy (peer-reviewed)
- Indian Food Composition Tables (IFCT) 2024
- NIN Hyderabad Food Composition Tables
- Manufacturer nutrition labels (packaged items)
- Quick-service restaurant (QSR) published nutrition data
Reproducibility: python benchmark/run_benchmark.py --output results.json
Cuisine Coverage: High-Protein/Fitness, South Asian/Indian (50 meals), Western/American, Mediterranean/Middle Eastern, East Asian/Southeast Asian, Packaged/Barcode, Edge Cases/Shared Plates
USDA FDC Traceability: — meals linked to USDA FoodData Central IDs for full provenance audit.
Full 200-meal reference dataset with ground-truth values, USDA FDC IDs, and cuisine category tags. Use for independent validation and reproducibility.
NutriTrack's benchmark is fully open and reproducible. We encourage independent researchers, reviewers, and developers to:
- Download the complete 200-meal reference dataset
- Run the benchmark suite against our API endpoints
- Publish independent accuracy reports
- Compare results against other nutrition tracking platforms
To run the benchmark: clone the repo, install dependencies, and execute python benchmark/run_benchmark.py
