Inspiration
Agriculture is the backbone of our economy, yet many farmers still rely on manual inspection to detect plant diseases. Cucumber crops are highly vulnerable to leaf infections, and delayed identification can lead to serious yield loss.
This inspired me to build an AI-powered solution that can detect diseases early using simple leaf images, making intelligent technology accessible and practical for farmers.
What It Does
This project uses Deep Learning to automatically detect and classify diseases in cucumber leaves from images.
The system:
Takes a leaf image as input
Extracts visual features automatically
Classifies the leaf as healthy or diseased
Displays the predicted disease category
It provides fast, accurate, and scalable crop disease diagnosis.
How I Built It Dataset
Created a custom dataset of cucumber leaf images
Included healthy and multiple disease categories
Cleaned and labeled images for supervised learning
Preprocessing
Resized images to 224 × 224
Normalized pixel values
Applied data augmentation (rotation, flipping, zooming)
Model Development
I first built a Custom 2D Convolutional Neural Network consisting of:
Convolution layers
ReLU activation
Max pooling layers
Fully connected layers
Softmax output layer
Later, I implemented EfficientNet to improve feature extraction and overall performance.
Results
92% accuracy using Custom CNN
95% accuracy using EfficientNet
Strong Precision, Recall, and F1-Score
Clear class separation in the confusion matrix
The final model showed reliable performance in distinguishing between healthy and diseased cucumber leaves.
What I Learned
Through this project, I learned:
Practical implementation of Deep Learning in real-world problems
Importance of high-quality and balanced datasets
Image preprocessing and augmentation techniques
Model tuning and performance optimization
Evaluating models using multiple performance metrics
This project strengthened my understanding of applying Artificial Intelligence to agriculture and building impactful, real-world solutions.
Future Improvements
Deploy as a mobile application for real-time leaf scanning
Expand dataset with additional disease categories
Provide treatment recommendations after detection
Integrate cloud-based monitoring for smart farming systems
Log in or sign up for Devpost to join the conversation.