CropSence

Inspiration

The inspiration for CropSense emerged from the persistent struggles of smallholder farmers in Bihar, India, where waterlogging after monsoons often delays wheat sowing, exposing crops to high-temperature stress during grain-filling and reducing yields by up to 20-30% in severe cases. Traditional satellite models merely classify crops like wheat, overlooking intercropping with mustard—a common practice to maximize land use amid water shortages and climate variability—leading to inaccurate insurance claims and subsidy allocations. Witnessing these issues during rural visits, combined with the transformative potential of satellite imagery in Indian agriculture—such as NASA's NISAR mission for real-time crop health monitoring and soil moisture tracking—I aimed to build a tool that turns raw data into actionable insights, empowering farmers, preventing fraud, and supporting policy in fragmented farms.

Built With

• 1d-cnn-(tensorflow/pytorch)
• agrifieldnet
• folium
• google-earth-engine
• python
• sentinel-1-api
• sentinel-2-api
• streamlit
• transformer
• xgboost