Inspiration
Agriculture is one of the largest industries in the world and forms the foundation of our food supply. However, pests, climate change, and plant diseases can severely damage crops. Many farmers struggle to detect and address these issues promptly. This project is our small step toward solving that problem.
What it does
Just take a picture of a potato plant's leaf and upload it. Our model will detect whether the plant has a disease (early blight, late blight) or is healthy. But there's more! We also provide an AI-based response explaining the cause of the issue and the recommended measures to take.
How we built it
We developed a CNN to determine if the plant is healthy or diseased. This prediction helps us generate AI-based advice on causes and treatments. We built a Flask app to make this accessible from almost any device. To avoid overfitting and improve validation accuracy, we implemented a learning rate scheduler and early stopping. We also used the Adam optimizer with adjusted beta_1 and beta_2 parameters to handle the challenge of recognizing patterns in complex data.
Challenges we ran into
The limited and complex data made it tough to identify patterns. However, our approach with the CNN model helped us overcome this challenge. We also planned to extend disease detection to other plants but couldn't complete it in time. We'll add this feature later.
Accomplishments that we're proud of
We achieved impressive accuracy, over 90%, on both test and training data. We successfully developed the Flask app and ensured everything works as expected.
What we learned
We gained knowledge about plant diseases, transfer learning, and how to fine-tune various parameters in Keras models to enhance performance and accuracy.
What's next for it
We aim to expand the system to detect diseases in other plants and improve our dataset by collecting higher-quality data in greater quantities.
Log in or sign up for Devpost to join the conversation.