AgriSens

• 

  AGRI-SENS-CORE platform homepage featuring target region controls and multi-source satellite data resolution settings.

• 

  Detailed climate anomaly and vegetation tabular view flagging drought and heat stress risks.

• 

  Soil moisture and land surface temperature anomaly z-score diagnostic charts highlighting stress thresholds.

• 

  Sentinel-2 NDVI time-series chart tracking canopy health dynamics over the user-selected date range against historical baseline averages.

• 

  System architecture documentation showing the end-to-end data ingestion, preprocessing pipeline, and CARI risk mathematical formulation.

• 

  Data export hub enabling downloads of time-series datasets (CSV), spatial boundaries (GeoJSON), and automated policy briefs.

• 

  Actionable remediation panel providing specific, calculated fertilizer and soil input proportions for high-risk fields.

• 

  High-resolution interactive field map displaying spatial AOI boundaries overlaid on Esri World Imagery.

• 

  Real-time field risk summary panel displaying calculated CARI index scores and sub-index metrics.

• 

  Date Range Settings panel allowing users to configure custom start and end dates along with satellite cloud cover percentage thresholds.

Inspiration

A lot of small-scale farmers still have to rely on guesswork when it comes to soil nutrients, crop selection, or diagnosing plant diseases. Catching an infection or planting the wrong crop for a specific soil type can wipe out a season's income. I created AgriSens to give farmers a simple, AI-powered toolkit that turns complex agricultural and environmental data into actionable advice they can trust.

What it does

AgriSens gives users a clear view of field conditions using satellite data and automated calculations:

Vegetation Health Tracking (NDVI): Processes multi-spectral satellite imagery to monitor crop canopy health and density over time using Normalized Difference Vegetation Index (NDVI) calculations.

Thermal & Environmental Anomaly Detection: Analyzes thermal band feeds and climate variables to spot sudden moisture stress, micro-climate changes, or land degradation early.

Interactive Visualization Dashboard: Translates complex geospatial raster files and remote sensing indices into intuitive visual maps, time-series graphs, and actionable reports.

How we built it

I built AgriSens using a custom Python pipeline for remote sensing and geospatial data: Geospatial Processing: Wrote scripts using Python geospatial libraries to process multi-spectral satellite imagery, extract band values, and run index calculations like NDVI. Data Processing: Used Pandas and NumPy to structure multi-dimensional raster arrays, run anomaly detection logic, and clean up time-series data. Dashboard: Built a fast web interface with Streamlit (supported by a FastAPI backend) to render field maps and performance charts smoothly

Challenges we ran into

Working with satellite raster files was brutal on memory. Loading high-resolution imagery caused major performance lag on the web dashboard early on, so I had to restructure how I process and slice raster layers before rendering. Fine-tuning the anomaly detection logic was also tricky—I had to make sure normal seasonal shifts in vegetation weren't incorrectly flagged as crop stress.

Accomplishments that we're proud of

I built a fully functional processing pipeline from scratch that converts complex satellite raster files into clear visual maps and automated crop health metrics.

What we learned

Building AgriSens taught me how to work with geospatial datasets, perform multi-spectral band math, and handle memory-heavy files in Python. I got much better at optimizing data processing pipelines and creating intuitive dashboards for complex scientific data.

What's next for AgriSens

I plan to connect AgriSens directly to Sentinel and Landsat APIs to pull new satellite imagery automatically and train predictive machine learning models for yield forecasting.

Built With

• data
• fastapi
• geospatial-data
• machine-learning
• ndvi
• numpy
• pandas
• python
• remote-sensing
• streamlit