Inspiration
I was inspired by the challenges faced by smallholder farmers who often lose significant portions of their harvest to diseases that could have been treated if caught early. I wanted to use my skills in Python and Machine Learning to build a tool that makes expert-level agricultural knowledge accessible to anyone with a smartphone.
What it does
Our application allows users to upload an image of a crop leaf. The system uses a Deep Learning (CNN) model to analyze the image and predict the specific disease (or confirm if it's healthy) with high accuracy. Beyond just prediction, we integrated Generative AI (Google Gemini) to act as a "Smart Doctor". It provides:
1.Diagnosis: Identifies the disease. 2.Treatment Plan: Recommends specific pesticides or natural remedies. 3.Prevention: Suggests farming practices to prevent future outbreaks.
How we built it
Frontend: Streamlit for a responsive and user-friendly interface. AI Model: Trained a Convolutional Neural Network (CNN) using TensorFlow/Keras on the PlantVillage dataset. AI Consultant: Integrated Google Gemini API to generate natural language advice based on the model's prediction. Backend: Python.
Challenges we ran into
Getting accurate predictions for similar-looking diseases. Integrating the heavy Deep Learning model with a lightweight web interface. Formatting the AI's response to be actionable and easy for farmers to read.
Accomplishments that we're proud of:
Achieving a high accuracy rate with our custom CNN model. Successfully combining traditional Computer Vision with modern LLMs (Gemini) for a complete solution. Creating a tool that has real social impact potential.
What's next for Crop-Disease-Prediction
Mobile App development for offline usage in remote fields. Multilingual support to help farmers in different regions. Community features for farmers to share insights.
Log in or sign up for Devpost to join the conversation.