AgroPulse AI

A Smart Streamlit-powered tool that uses Gemini AI to detect plant diseases from leaf images and delivers instant, actionable cures with downloadable PDF reports.

Comment

Inspiration

Farmers often face difficulties identifying plant diseases early and lack easy access to personalized weather insights, leading to crop damage and financial loss. We wanted to build a scalable AI-powered solution that helps farmers detect plant diseases just by uploading a leaf image and get weather predictions — all in real-time without maintaining any servers.

What it does

AgroPulse AI is a Gemini API-powered tool that allows farmers to:

📸 Detect plant diseases from uploaded leaf images using multimodal AI (Gemini 1.5 Flash)

🧪 Get smart cure suggestions based on the AI diagnosis

📂 Download PDF reports summarizing disease, cure, and forecast

It’s a fully serverless architecture with Streamlit UI and real-time AI backend integration.

How we built it

We built AgroPulse AI using:

Python + Streamlit for the frontend and user interactions

Gemini 1.5 Flash API for real-time multimodal disease detection and language generation

AWS Lambda to process AI inference tasks (PDF generation, prompt handling)

AWS API Gateway to trigger Lambda from Streamlit

Optional: Amazon S3 for storing images temporarily

OpenWeatherMap API for fetching and visualizing temperature + precipitation forecasts

PDFKit + Jinja2 for PDF report generation

Challenges we ran into

Integrating image input with Gemini’s multimodal API

Designing prompt engineering to generate accurate, actionable cures

Keeping serverless Lambda functions lightweight and fast

Handling multiple image uploads and rendering forecasts cleanly

Accomplishments that we're proud of

Successfully integrated Gemini 1.5 Flash for real-time plant disease detection

Designed an intuitive and futuristic Streamlit UI

Generated PDF reports dynamically with disease + weather insights

Deployed a serverless AI workflow using Lambda + API Gateway

What we learned

Practical prompt design for multimodal Gemini APIs

Real-world integration of serverless architecture with frontend tools like Streamlit

Optimizing API request/response pipelines for performance and cost

Creating scalable agri-tech solutions that can be production-ready

What's next for AgroPulse AI

Integrate voice support and regional language output

Expand disease detection support to cover more crops and regions

Build a mobile version of the tool for offline use

Add crop health trend analysis and alert systems

Built With

  • ai-powered-diagnosis
  • fpdf
  • geminiapi
  • pillow
  • python
  • regex
  • streamlit
Share this project:

Updates

posted an update —

I have made some major improvements to AgroPulse AI.

Multiple Image Upload – Users can now upload several plant leaf images at once. Each image gets processed individually with separate disease predictions.

Multiple Disease Predictions per Image – Our app now supports detecting more than one disease in a single leaf image, improving diagnostic accuracy.

PDF Report Generation – After analysis, a downloadable report is generated containing disease details, suggested cures, and a summary of the user's input.

Powered by Gemini 1.5 Flash API for fast and intelligent responses.

Live Demo: https://agropulse-ai.streamlit.app/ Source: GitHub Repo

Log in or sign up for Devpost to join the conversation.