Back to work
Autonomous Hardware & UAVJune 2026

Drishti — Precision-Agriculture Ground Station

Drone imagery in, crop-health maps and spray prescriptions out.

4
Crop-health zones
3
Vegetation indices

Overview

A Windows ground station for an autonomous hexacopter. It takes paired RGB and near-infrared images from two Raspberry Pi cameras, computes vegetation indices (NDVI, SAVI, GNDVI), classifies crop health into four zones, and exports spray-prescription maps. In active development.

The Problem

Spraying a whole field evenly wastes pesticide and fertiliser on healthy crops. Targeting only the stressed areas needs current, reliable maps of crop health.

Approach

Paired RGB and NIR frames are aligned, converted to vegetation indices, classified into four health zones by NDVI thresholds, and exported as Shapefile or GeoJSON prescriptions. The desktop app (PySide6) has FLY, SCAN, HISTORY and SETTINGS pages, reads flight telemetry over MAVLink, and keeps field and flight history in SQLite.

Hardware

Hexacopter airframe
Raspberry Pi 4 with NoIR camera (near-infrared)
Raspberry Pi Zero 2 W with RGB camera

Highlights

Band extraction, vegetation indices and four-zone classification working end to end on synthetic data, with an automated test suite.
Modular pipeline: ingest, pairing, alignment, indices, orthomosaic, classification, prescription export and viewer.

Technologies & Tools

PythonPySide6OpenCVrasterioGeoPandasMAVLinkSQLiteOpenDroneMap