Machine Learning

This repository contains machine learning models and implementations for the StunBy application, which focuses on combating stunting through innovative technology.

Table of Contents

• Baby Height Detection
• Nutrition Prediction
• Z-Score Calculation & Classification
• Food Recommendation & Tracking
• Chatbot with Vertex AI & RAG (Retrieval Augmented Generation)
• API Machine Learning Documentation
  • How to Run
  • Measure Classify
  • Predict Nutrition
  • Add Food Tracking
  • Recommendation Food
• Contributors

Computer Vision to Baby Height Measurement Detection

• Model: Fine-tuned YOLO on COCO dataset (people and coin classes).
• Dataset: Coco dataset Coin dataset roboflow
• Notebook: Notebook People detection Notebook Coin detection
• Purpose: Detect baby's height from images with reference to a coin for scale.
• Input: Vertical images capturing the baby from head to toe with a coin visible.
• Output: Estimated height of the baby in centimeters.
• Dependencies:
  • TensorFlow
  • OpenCV
  • ultralytics
  • YOLOv11

Nutrition Prediction

• Model: TensorFlow-based model for predicting nutrient requirements.

• Dataset: WHO child growth standards dataset

• Notebook: Notebook

• Purpose: Calculate the daily caloric, protein, fat, and carbohydrate needs of a baby.

• Input: Age, gender, weight, height, activity level, and breastfeeding status.

• Output: Recommended daily intake of nutrients.

• Dependencies:

  • TensorFlow
  • NumPy
  • Pandas
  • Scikit-learn

• Model Evaluation:
  

  

Z-Score Calculation & Classification

• Purpose: Classify the child's growth status based on WHO Z-Score standards.
• Dataset: Kementerian Kesehatan RI
• Notebook: Notebook
• Input: Baby's height, weight, and age.
• Output: Growth classification (e.g., normal, stunted).
• Dependencies:
  • NumPy
  • WHO child growth standards dataset

Food Recommendation & Tracking

• Model: Content-Based and Collaborative Filtering.
• Dataset: Baby food dataset
• Notebook: Notebook
• Purpose: Recommend suitable baby foods and track daily food intake.
• Input: Baby's dietary preferences, allergies, and nutrient needs.
• Output: Personalized food recommendations and intake tracking.
• Dependencies:
  • Scikit-Learn
  • Pandas
  • Surprise (Collaborative Filtering library)

Chatbot with Vertex AI & RAG (Retrieval Augmented Generation)

• Model: Retrieval-Augmented Generation using Vertex AI.
• Notebook: Notebook
• Purpose: Provide quick responses to parenting and nutrition-related queries.
• Input: User queries about stunting, nutrition, or app features.
• Output: Conversational responses enriched with relevant information.
• Dependencies:
  • Vertex AI SDK
  • Flask (for API wrapping)

API Machine Learning Documentation

How to Run

1. Clone this repository.
2. Go to the ML_API_Integration directory.
3. Install Dependencies

pip install -r requirements.txt

4. Run the API

python app.py

Measure Classify

Untuk mengukur panjang badan dan mengklasifikasikan status gizi berdasarkan panjang badan, berat badan, umur, dan jenis kelamin.

POST http://127.0.0.1:5000/measure-classify Request:

{
  "url": "https://storage.googleapis.com/stunby_bucket/baby_testing/baby_3.jpeg",
  "weight": 5.5,
  "age": 6,
  "gender": "male"
}

Response:

{
  "baby_length": 45.79,
  "imt": 25.99243856332703,
  "nutritional_status_length": "Sangat Pendek", // "Sangat Pendek", "Pendek", "Normal", "Tinggi", "Tidak Diketahui"
  "nutritional_status_weight": "Gizi Kurang", // "Gizi Buruk", "Gizi Kurang", "Gizi Baik", "Gizi Lebih", "Kategori Tidak DIketahui"
  "status_bb_tb": "Gemuk", // "Sangat Kurus", "Kurus", "Normal", "Gemuk", "Kategori Tidak Diketahui"
  "status_imt": "Obesitas (Obese)", // "Gizi Buruk (Severely Wasted)", "Gizi Kurang (Wasted)", "Gizi Baik (Normal)", "Berisiko Gizi Lebih (Possible Risk of Overweight)", "Gizi Lebih (Overweight)", "Obesitas (Obese)", "Kategori Tidak Diketahui"
  "z_score_bb_tb": 9.666666666666671,
  "z_score_length": -10.285714285714311,
  "z_score_weight": -2.999999999999998
}

Predict Nutrition

Untuk memprediksi kebutuhan nutrisi berdasarkan umur, jenis kelamin, berat badan, tinggi badan, tingkat aktivitas, dan status pemberian ASI.

POST 127.0.0.1:5000/predict_nutrition request:

{
    "usia_bulan": 12,           # int: 0-24 bulan
    "gender": "L",              # str: "L" atau "P"
    "berat_kg": 9.5,           # float: berat dalam kg
    "tinggi_cm": 75.0,         # float: tinggi dalam cm
    "aktivitas_level": "Sedang", # str: "Rendah"/"Sedang"/"Aktif"/"Sangat_Aktif"
    "status_asi": "ASI+MPASI"   # str: "ASI_Eksklusif"/"ASI+MPASI"/"MPASI"
}

response:

{
  "status": "success",
  "data": {
    "calories_needed": 850.5, // kalori dalam kkal
    "proteins_needed": 20.1, // protein dalam gram
    "fat_needed": 30.2, // lemak dalam gram
    "carbohydrate_needed": 120.5 // karbohidrat dalam gram
  }
}

Add Food Tracking

Untuk menambahkan makanan yang dikonsumsi oleh anak.

POST http://127.0.0.1:5000/api/tracking/add-food

request:

{
  "food_name": "ASI (Air Susu Ibu)",
  "portion": 200
}

response:

{
  "data": {
    "name": "ASI (Air Susu Ibu)",
    "notes": "ASI Eksklusif direkomendasikan oleh WHO sebagai sumber nutrisi eksklusif.",
    "nutrients": {
      "calcium": 68.0,
      "calories": 140.0,
      "carbohydrate": 14.0,
      "fat": 8.4,
      "proteins": 2.4
    },
    "portion": 200.0
  },
  "status": "success"
}

recommendation food

Untuk merekomendasikan makanan yang sesuai dengan kebutuhan nutrisi anak.

POST http://localhost:5000/recommend-food

REQUEST

{
  "age_months": 3,
  "daily_needs": {
    "calorie": 450,
    "protein": 9.5,
    "carb": 31,
    "fat": 48
  },
  "daily_budget": 50000,
  "user_preferences": ["ASI"] // optional, bisa berupa kata kunci yang diinginkan
}

{
  "age_months": 7,
  "daily_needs": {
    "calorie": 700.5,
    "protein": 20.1,
    "carb": 120.5,
    "fat": 30.2
  },
  "daily_budget": 30000,
  "user_preferences": ["Susu", "Buah"] // optional, bisa berupa kata kunci yang diinginkan
}

RESPONSE

{
  "recommendations": [
    {
      "Banyak_produk": 1.0,
      "Berat_per_Produk(gr)": 150,
      "Harga_per_Porsi(IDR)": 15800.000000000002,
      "Kalori_per_Porsi(kcal)": 108.0,
      "Karbohidrat_per_Porsi(gr)": 12.0,
      "Lemak_per_Porsi(gr)": 5.700000000000001,
      "Nama_Makanan": "Susu Formula Lanjutan 6-12 bulan",
      "Notes": "Contoh merek: Bebelac Gold 2 SGM 2 Dancow 2. Diperkaya dengan zat besi dan vitamin esensial.",
      "Protein_per_Porsi(gr)": 2.6999999999999997,
      "Total_Harga": 15800.000000000002
    },
    {
      "Banyak_produk": 2.0,
      "Berat_per_Produk(gr)": 100,
      "Harga_per_Porsi(IDR)": 8000.0,
      "Kalori_per_Porsi(kcal)": 160.0,
      "Karbohidrat_per_Porsi(gr)": 8.5,
      "Lemak_per_Porsi(gr)": 14.699999999999996,
      "Nama_Makanan": "Pure Alpukat",
      "Notes": "Alpukat matang dilumatkan. Sumber lemak baik untuk perkembangan otak bayi.",
      "Protein_per_Porsi(gr)": 2.0,
      "Total_Harga": 16000.0
    },
    {
      "Banyak_produk": 1.0,
      "Berat_per_Produk(gr)": 500,
      "Harga_per_Porsi(IDR)": 10000.0,
      "Kalori_per_Porsi(kcal)": 384.99999999999994,
      "Karbohidrat_per_Porsi(gr)": 85.0,
      "Lemak_per_Porsi(gr)": 0.5,
      "Nama_Makanan": "Pure Kentang",
      "Notes": "Kentang kukus yang dihaluskan. Sumber karbohidrat yang baik.",
      "Protein_per_Porsi(gr)": 10.0,
      "Total_Harga": 10000.0
    }
  ],
  "summary": {
    "remaining_budget": -11800.0,
    "total_nutrients": {
      "Kalori": 813.0,
      "Karbohidrat": 114.0,
      "Lemak": 35.599999999999994,
      "Protein": 16.7
    }
  }
}

{
  "error": "Tidak ada solusi yang ditemukan. Coba lagi dengan budget yang lebih tinggi."
}

Contributors

• Ridlo Abdullah Ulinnuha
• Ahmad Dzulfikar Ubaidillah
• Azzahra Athifah Dzaki