🧼 The IoT Hygiene Ecosystem - AI Model Infrastructure

Transforming Public Sanitation through Real-Time Monitoring & Digital Trust

This directory contains the complete AI model infrastructure for the Smart Toilet Hygiene Monitoring System, designed to restore public trust in toilets using IoT sensors, real-time hygiene scoring, and automated sanitation actions.

🚀 Project Overview

The IoT Hygiene Ecosystem is a smart sanitation solution targeting urban, rural, and institutional public toilets under India's Swachh Bharat Mission (SBM) framework.

📅 Target Deployment: July 2025
🎯 Theme: Smart Cities | IoT | AI | Public Health | Digital India

❗ Problem Statement: The Sanitation Paradox (2025)

India has successfully built toilets, but people still don't trust them.

✅ 95%+ villages are ODF Plus certified
❌ Only 10% of users trust public toilet maintenance
🚫 68% of people avoid public toilets entirely
🧠 Root cause: Psychological Scarring due to bad odor, dirty floors, no water, and unsafe environments

Result: Infrastructure exists, but usage does not.

💡 Our Solution

We convert a public toilet into a "Living Smart Unit" that:

Detects hygiene issues before users arrive
Automatically responds using IoT-driven actions
Displays a real-time Hygiene Trust Score outside the toilet
Rebuilds confidence through pre-entry validation

🔍 Key Features

🧪 Real-Time Hygiene Monitoring

Ammonia & odor detection
Water and soap level tracking
Touchless usage detection

📊 Digital Trust Score

Live hygiene score display
Last cleaned / disinfected time
QR-based hygiene verification for users & tourists

🤖 Automated Sanitation

Auto exhaust & odor neutralization
UV-C sterilization after usage
Touchless flush and taps

🚨 Safety & Inclusion

Panic/help button
Slip detection (future scope)
Women & elderly-friendly design

🧠 Behavioral Research (User-Centric Design)

We conducted a 10-point questionnaire to identify real user pain points:

Fear of bad smell & dirty floors
Avoidance of water intake
Surface-touch anxiety
Willingness to pay for guaranteed hygiene
Trust in digital hygiene scores

📌 Insight: Smell and first impression decide usage within 3 seconds

🧩 IoT Technical Architecture

🔌 Sensors Used

Sensor	Purpose
MQ-135 / MQ-137	Ammonia & odor detection
PIR Sensor	User presence detection
Ultrasonic Sensor	Water tank level
Flow Sensor	Water & soap usage
UV-C Module	Post-use sterilization

⚙️ Automated Actions

High-speed exhaust activation
Ozone / odor neutralization
UV sterilization after exit
Alert generation for maintenance staff

🌍 Sector-Specific Deployment

🏡 Rural (SBM-Gramin)

Solar-powered IoT
LoRaWAN communication
Low-internet dependency

🏙️ Urban (Malls, Parks, Smart Cities)

Premium "Loo-Café" hygiene model
Pay-per-use with guaranteed cleanliness

🏫 Schools & Colleges

Soap-usage tracking
Handwashing compliance monitoring
Child-safe hygiene assurance

🧪 Innovation Highlights

Pre-Entry Hygiene Validation
Psychological Barrier Reduction
Global Hygiene QR for Tourists
Behavior-first sanitation design

📈 Impact

Increased public toilet usage
Reduced health risks & infections
Higher citizen satisfaction scores
Data-driven sanitation governance
Alignment with Super Swachh League (2025) metrics

🔮 Future Scope

AI-based hygiene prediction
Computer vision for cleanliness detection
Mobile app for hygiene navigation
Government dashboard integration
Predictive maintenance alerts

🏁 Conclusion

Public sanitation in 2025 requires more than infrastructure — it requires Digital Trust.

The IoT Hygiene Ecosystem bridges the gap between:

"There is a toilet" and
"I feel safe using this toilet."

📁 AI Model Infrastructure Directory Structure

Directory Structure

ai_model/
├── data/                    # Data storage and processing
│   ├── raw/                # Raw sensor data from IoT devices
│   ├── processed/          # Cleaned and preprocessed data
│   └── synthetic/          # Augmented and synthetic data
├── models/                 # Trained ML models and artifacts
├── training/               # Model training infrastructure
│   ├── scripts/           # Training scripts and notebooks
│   └── logs/              # Training logs and metrics
├── testing/               # Testing and validation
│   ├── unit_tests/        # Unit tests for model components
│   └── integration_tests/ # End-to-end pipeline tests
├── evaluation/            # Model evaluation and performance metrics
├── utils/                 # Utility functions and helpers
└── config/                # Configuration files and parameters

## Quick Start

### 1. Training the Model
```bash
cd training/scripts
python train_model.py

2. Running Tests

# Unit tests
cd testing/unit_tests
python test_model.py

# Integration tests
cd testing/integration_tests
python test_pipeline.py

3. Data Flow

Raw Data Collection: IoT sensors collect environmental data
Data Processing: Clean and preprocess sensor readings
Model Training: Train hygiene prediction models
Model Testing: Validate model performance
Deployment: Deploy trained models to production

Data Format

Training Features

humidity: Relative humidity percentage (0-100)
temperature: Temperature in Celsius (15-35°C)
ammonia_level: Ammonia concentration (ppm)
co2_level: CO2 concentration (ppm)
occupancy_duration: Average occupancy time (seconds)
usage_count: Number of toilet uses
cleaning_frequency: Cleaning frequency per day
hygiene_score: Target variable (0-100)

Testing Features

Same as training features but without the target variable (hygiene_score).

Model Architecture

The current implementation uses:

Algorithm: Random Forest Regressor
Features: 7 environmental and usage metrics
Target: Hygiene score (0-100)
Evaluation: MSE, R² Score

Performance Metrics

MSE: Mean Squared Error
R² Score: Coefficient of determination
Feature Importance: Ranking of input features

File Descriptions

Data Files

sensor_readings.csv: Raw sensor data from IoT devices
training_features.csv: Preprocessed training data
testing_features.csv: Preprocessed testing data
augmented_data.csv: Synthetic and augmented training data

Training Scripts

train_model.py: Main training script with model pipeline

Test Files

test_model.py: Unit tests for model functionality
test_pipeline.py: Integration tests for data pipeline

Usage Examples

Loading and Using the Model

import joblib
import pandas as pd

# Load trained model
model = joblib.load('models/hygiene_predictor.pkl')

# Prepare input data
input_data = pd.DataFrame({
    'humidity': [65.0],
    'temperature': [22.0],
    'ammonia_level': [0.8],
    'co2_level': [450],
    'occupancy_duration': [180],
    'usage_count': [12],
    'cleaning_frequency': [8]
})

# Make prediction
hygiene_score = model.predict(input_data)
print(f"Predicted hygiene score: {hygiene_score[0]:.1f}")

Data Preprocessing

import pandas as pd

def preprocess_sensor_data(raw_data):
    """Preprocess raw sensor data for model input"""
    # Handle missing values
    processed_data = raw_data.fillna(method='forward_fill')
    
    # Normalize features if needed
    # Add feature engineering logic here
    
    return processed_data

Configuration

Model configuration files are stored in the config/ directory:

Model hyperparameters
Feature engineering settings
Evaluation metrics configuration

Testing Strategy

Unit Tests: Test individual model components
Integration Tests: Test complete data pipeline
Performance Tests: Monitor model performance metrics
Data Quality Tests: Validate input data integrity

Future Enhancements

Deep learning models (LSTM, CNN)
Real-time model updates
Advanced feature engineering
Model explainability tools
Automated hyperparameter tuning
Model versioning and A/B testing

Dependencies

pandas
numpy
scikit-learn
joblib
pytest (for testing)

Contributing

Add new training data to data/raw/
Update preprocessing scripts in training/scripts/
Run tests to ensure model quality
Update documentation and metrics

License

This AI model is part of the Smart Toilet Hygiene Monitoring System.

FilesExpand file tree

README.md

Latest commit

History