This repository contains the implementation of an AI-Controlled Automated Hydroponic Farming System, developed as part of our interdisciplinary engineering project.
The system integrates IoT sensors, embedded controllers, and AI-based plant health monitoring to create a self-regulating hydroponic farm.
- Automatically monitors temperature, humidity, water level, TDS, moisture, and UV exposure.
- Controls pumps, fans, and lighting using sensor feedback.
- Detects plant diseases in real-time using a VGG16 CNN model with ~95% accuracy.
- Provides alerts and dashboard updates through IoT platforms (Blynk/Telegram).
This project demonstrates how AI + IoT + Hydroponics can address challenges of sustainable agriculture, urban farming, and efficient resource utilization.
| Component | Purpose |
|---|---|
| ESP32 | Central microcontroller for sensor data acquisition and automation |
| DHT22 Sensor | Measures ambient temperature & humidity |
| DS18B20 Sensor | Monitors water temperature in nutrient solution |
| TDS Sensor | Measures nutrient concentration (ppm) |
| Moisture Sensor | Detects root-zone moisture in grow medium |
| Water Level Sensor | Prevents dry reservoir conditions |
| UV Sensor | Tracks light intensity exposure |
| Water Pump | Circulates nutrient solution |
| Inlet/Exhaust Fans | Maintains air circulation & temperature control |
| Relay Modules | Controls actuators safely |
| Power Supply | 5V/12V regulated supply for MCU and actuators |
-
Programming Languages:
- Arduino C/C++ (for ESP32 firmware)
- Python (for AI model & image processing)
-
AI / ML Frameworks:
- TensorFlow & Keras (VGG16 training & deployment)
- OpenCV (image preprocessing, camera interface)
- NumPy / Pandas (data handling)
-
IoT Platforms:
- Blynk Cloud (real-time dashboard)
- Telegram Bot API (alerts & notifications)
-
Development Tools:
- Arduino IDE (firmware upload)
- VS Code / Jupyter Notebook (AI development)
- Proteus / Tinkercad (circuit simulation)
-
Hardware Setup
- Assemble ESP32 with sensors & actuators.
- Connect sensors via GPIO/ADC pins.
- Power actuators using relay modules.
- Place sensors in hydroponic setup (water tank, root zone, canopy).
-
Firmware Development (ESP32)
- Collects sensor readings at intervals.
- Implements threshold-based actuator logic (e.g., pump ON if TDS < 700 ppm).
- Sends sensor data to cloud (Blynk/Telegram).
-
AI Model (Plant Disease Detection)
- Dataset: PlantVillage + custom images (~54,000 samples).
- Architecture: VGG16 CNN (pruned for edge deployment).
- Accuracy: 96.4% (training), 93.8% (validation).
- Real-time inference speed: ~1.2s per image.
-
System Integration
- ESP32 handles automation loop.
- AI module (Python) runs on Raspberry Pi/PC with camera feed.
- IoT dashboard visualizes live data & alerts.
| Pump & Wiring Setup | ESP32 Pinout | Circuit Assembly |
|---|---|---|
 |
 |
 |
-
Automation Performance
- Stable maintenance of optimal hydroponic conditions.
- Actuators triggered in real-time based on thresholds.
-
AI Disease Detection
- Identified diseases like leaf spot, blight with high accuracy.
- Live alerts sent with disease label & leaf image.
-
System Reliability
- Successfully tested for 48–72 hours continuous operation.
- Sensor calibration ensured ±2% accuracy.
| Final Model (Daylight) | Final Model (Night Mode) |
|---|---|
 |
 |
- Expand dataset for multi-crop disease detection.
- Deploy lightweight AI models on ESP32-CAM (edge computing).
- Integrate mobile app for farmers with offline alerts.
- Enable nutrient dosing automation using peristaltic pumps.
- Scale prototype to commercial hydroponic greenhouses.
Additional Resources
📑 The full LaTeX project report, IEEE paper, and presentation (PPT) are available.
📧 Contact for more info: