Edge-AI Crop Disease Classifier & Precision Irrigation Hub

Farmers and growers often rely on manual crop inspection and fixed irrigation schedules, which can delay disease identification and lead to unnecessary or insufficient watering.

This project demonstrates how computer vision, soil sensing and automated irrigation can be combined to support faster crop-condition assessment and more targeted water management.

1. The ESP32-CAM captures an image of the crop leaf and sends it to the configured Edge-AI classification process.
2. A trained lightweight image-classification model analyzes the leaf and identifies whether it matches one of the supported healthy or disease classes.
3. The soil probe measures available parameters such as soil nutrient levels and other supported soil conditions, depending on the selected sensor module.
4. The ESP32-S3 processes sensor readings and compares them with predefined irrigation thresholds.
5. When irrigation is required, the controller activates the corresponding relay channel and opens the required solenoid valve.
6. Different irrigation zones can be controlled independently based on the configured prototype logic.
7. Wi-Fi can optionally be used for dashboard monitoring, historical logging or remote alerts.

1. AI-based crop leaf disease classification from camera images
2. ESP32-CAM-based leaf image capture for plant monitoring
3. Soil nutrient and condition monitoring using an NPK sensor
4. Automatic irrigation control based on configured soil thresholds
5. Multi-zone irrigation using relay-controlled solenoid valves
6. ESP32-S3-based sensor processing and Edge-AI integration
7. Configurable irrigation and crop-monitoring parameters
8. Expandable Wi-Fi connectivity for remote monitoring and data logging

1. Source code
2. Circuit diagram
3. Component list
4. Project report
5. Abstract PDF
6. PPT / presentation
7. Installation guide
8. Setup support
9. Demonstration video
10. AI model integration and sensor calibration guidance

Prerequisites:

1. The disease-classification model must be trained or configured for the selected crop and supported disease classes.
2. Soil sensors and irrigation thresholds require proper installation, calibration and testing.

Limitations:

1. This is an academic smart-agriculture prototype and not a certified agricultural diagnostic system.
2. Disease-classification accuracy depends on the training dataset, leaf visibility, camera angle, lighting and image quality.
3. The AI model can identify only the crop diseases/classes included in its trained dataset.
4. Soil NPK and related sensor readings depend on sensor quality, calibration and soil conditions.
5. Automated irrigation decisions are based on configured thresholds and prototype sensor readings.
6. Wi-Fi/internet is required only for optional remote monitoring or cloud features.
7. Hardware components may vary depending on availability.
8. Custom dashboards, cloud hosting, mobile apps and additional sensors require separate customisation.