Multi-Pest Detection and Multispectral Imaging Payload for Toor Dal

Toor Dal, a staple crop in India, is highly susceptible to pests like the pod borer. Traditional visual inspection is slow and often happens after irreversible damage. Late detection leads to excessive, preventative pesticide spraying which harms the environment and reduces farmer profit margins.

This project brings cutting-edge multispectral imaging from the lab to the agricultural field. By attaching a custom-built, low-cost sensor array to commercial drones, the system can autonomously scan large crop areas and identify the specific spectral signatures of early-stage pest stress.

The lab team developed specialized optical filters that isolate wavelengths indicative of plant stress. This data is fed into an edge-computing AI model directly on the drone. Farmers receive a simple, color-coded heat map on their smartphones indicating precisely which sections of the field require attention.

The payload prototype has been successfully miniaturized and tested across three harvest cycles in Maharashtra. The AI model currently achieves an 88% accuracy rate in detecting stress signatures up to 10 days before visual symptoms appear.

By enabling targeted pesticide application, this technology can reduce chemical usage by up to 40%, lower farming costs, and protect local soil health, directly stabilizing the income of smallholder farmers.

Future iterations will integrate thermal imaging to simultaneously detect water stress. We are also working on a cloud-based dashboard for agricultural cooperatives to monitor regional pest movement patterns.