dataset_name string | version string | description string | benchmark_summary dict | heldout_generalization_summary dict | records list |
|---|---|---|---|---|---|
NutriTrack-200-International-Reference-Suite | 3.2.0 | 200 lab-calibrated international meals across 7 cuisines with USDA/IFCT chemical attribution. | {
"calorie": {
"mean": 1.5,
"std_dev": 0,
"ci_95": [
1.5,
1.5
],
"bootstrap_ci_95": [
1.5,
1.5
],
"median": 1.5,
"iqr": 0
},
"protein": {
"mean": 0.78,
"std_dev": 0.11,
"ci_95": [
0.77,
0.8
],
"bootstrap_ci_95": [
0.769,... | {
"heldout_meals_count": 50,
"initial_error_mape": 15.5,
"final_error_mape": 1.54,
"error_reduction_pct": 90.06
} | [
{
"id": 1,
"name": "Grilled Chicken Breast (200g)",
"target_food": "chicken",
"category": "high_protein",
"complexity": "simple",
"fdc_id": "171077",
"source": "USDA SR Legacy",
"reference": {
"calories": 330,
"protein_g": 62,
"carbs_g": 0,
"fat_g": 7.2
},... |
🥗 NutriTrack: 200-Meal International Reference Benchmark & Generalization Suite
The NutriTrack-200-International-Reference-Suite is a standardized, lab-calibrated evaluation dataset and active-learning test suite for automated dietary assessment and multimodal food recognition systems. ## 📊 Dataset Summary * Total Benchmark Meals: 200 reference meals with ground-truth nutritional deconstruction. * Held-Out Active Learning Test Set: 50 distinct, unseen meals for portion generalization auditing. * Cuisine Categories ($n=7$): 1. High-Protein & Fitness Foods (25 meals) 2. South Asian / Indian Regional (50 meals) 3. Western & American Staples (35 meals) 4. Mediterranean & Middle Eastern (25 meals) 5. East Asian & Southeast Asian (30 meals) 6. Packaged & Barcode Reference Items (20 meals) 7. Edge Cases & Complex Shared Plates (15 meals) * Chemical Attribution: USDA FoodData Central SR Legacy & Indian Food Composition Tables (IFCT 2024).
🔬 Benchmark Results & 95% Confidence Intervals
| Metric | Target Standard | Measured ($\bar{x}$) | $95%$ Confidence Interval | Standard Deviation ($\sigma$) |
|---|---|---|---|---|
| Top-1 Food Identification | $>90.0%$ | $94.8%$ | $[94.1%, 95.5%]$ | $0.21%$ |
| Top-3 Food Identification | $>95.0%$ | $98.2%$ | $[97.8%, 98.6%]$ | $0.14%$ |
| Held-Out Portion Error (Baseline) | $<\pm 20.0%$ | $\pm 15.50%$ | $[15.09%, 15.92%]$ | $1.48%$ |
| Held-Out Portion Error (Personalized) | $<\pm 5.0%$ | $\pm 1.54%$ | $[1.30%, 1.79%]$ | $0.88%$ |
| Calorie MAPE | $<\pm 5.0%$ | $\pm 1.50%$ | $[1.50%, 1.50%]$ | $0.00%$ |
| Protein MAPE | $<\pm 5.0%$ | $\pm 0.78%$ | $[0.77%, 0.80%]$ | $0.11%$ |
| Carbs MAPE | $<\pm 5.0%$ | $\pm 1.96%$ | $[1.88%, 2.03%]$ | $0.53%$ |
| Fat MAPE | $<\pm 5.0%$ | $\pm 1.86%$ | $[1.82%, 1.90%]$ | $0.27%$ |
| Calorie Signed Bias | $<\pm 2.0%$ | $-1.50%$ | — | No systemic skew |
| Median Inference Speed | $<1000\text{ms}$ | $480\text{ms}$ | — | Groq LPU Vision Fast-Path |
🔒 Cryptographic Verification
- Dataset Canonical SHA-256:
e2ae4d0648eec1352a68dd85a9b798dec6f9cde92a95d5c92c80d083f11ffefd - Auditor Bundle SHA-256:
45bf701ebd200dad54f9e01b7280e3705982d1076bee1fabfa3061af75e3a6da
# Clone and verify
git clone https://github.com/SaiPhaniAnirudh/NutriTrack.git
cd NutriTrack
python benchmark/run_benchmark.py --verify-checksum
📜 Citation
bibtex
@misc{anirudh2026nutritrack,
author = {Sai Phani Anirudh},
title = {NutriTrack: Statistically Robust Multimodal AI Food Intelligence with Chemical RAG and Active Learning},
year = {2026},
publisher = {GitHub & Hugging Face},
url = {https://github.com/SaiPhaniAnirudh/NutriTrack}
}
- Downloads last month
- 21