FinalistApril 2026Mehsana,Gujarat
Convergence 2026
KissanSahyog AI-powered platform for Farmers
Role: Machine Learning Developer
Project Overview
At our college hackathon, our team of 4 built KissanSahyog an AI-powered platform that helps Indian farmers make smarter decisions about their crops.
The idea was simple: a farmer shouldn't need to be a data scientist to know why their crops are failing or what to grow next season. So we built something that does the heavy lifting for them.
Here's what KissanSahyog can do:
📸 Take a photo of a diseased plant and get an instant diagnosis with a full treatment plan in English, Hindi, or Gujarati.
📊 Predict crop yield and explain exactly which factors (rainfall, soil nutrients, temperature) are responsible not just a number, but a reason.
🌱 Recommend the right crop and fertilizer for your land based on your actual soil composition.
🛰️ View your field's health from satellite imagery using live NDVI maps via Sentinel-Hub.
The best part? You only need to share your GPS location. Everything else soil data, weather, satellite readings is fetched automatically.
On the technical side, we built 4 custom ML models and open-sourced all of them on Hugging Face:
➢ ArogyaDrishti: Disease detection using ResNet50 + DenseNet121 (PyTorch)
➢ Praapthi: Yield prediction with Explainable AI (XGBoost)
➢ Bijamitra: Crop recommendation based on soil intelligence
➢ Poshan: Fertilizer guidance using real N-P-K analysis
Data comes from Google Earth Engine (SoilGrids), NASA POWER API, Open-Meteo, and Sentinel-Hub.
The stack: FastAPI · Next.js 15 · TailwindCSS · Supabase · Docker · Gemini AI