Methodology
The project follows an iterative design-science approach that combines system development with technical evaluation.
Research Design
The study is applied because it proposes a practical system, exploratory because it investigates the use of AI in organic gardening, and experimental because the machine-learning models are tested. Development proceeds through requirements analysis, data collection, data preparation, model creation, interface design, integration and evaluation.
Requirements
Identify user needs, gardening challenges and expected outputs.
Data Collection
Collect or simulate environmental data and obtain plant images.
Pre-processing
Clean readings, normalise values, resize images and create datasets.
Model Development
Develop irrigation and image-classification models.
Interface Design
Create a responsive website for results and recommendations.
Evaluation
Test system functions, model performance and usability.
Environmental Data
Soil moisture, temperature and humidity are the main variables used to estimate irrigation need. Data can be collected from physical sensors or simulated datasets when hardware is unavailable.
Plant Image Data
Healthy and diseased leaf images are prepared for classification. Images are resized, labelled and divided into training and testing sets.
Irrigation Prediction
Regression methods such as Linear Regression, Decision Tree Regressor and Random Forest Regressor can estimate watering requirements from environmental conditions.
Disease Classification
A Convolutional Neural Network can learn colour, texture and lesion patterns to classify plant images into healthy or disease-related categories.
Tools
| Technology | Purpose |
|---|---|
| Python and Scikit-learn | Data preparation, regression and evaluation. |
| TensorFlow/Keras | Deep-learning image classification. |
| OpenCV | Image preparation and enhancement. |
| HTML and CSS | Responsive website presentation. |
| IoT sensors | Environmental data collection. |