Forest fires can cause widespread damage to ecosystems and communities, while delays in detection give fires more time to spread. Current Earth-observation satellites can transmit large volumes of imagery to ground stations for analysis, a process that can delay fire detection while consuming limited bandwidth and power.
SASTRA Deemed to be University is developing a low-power VLSI-based AI chip designed to detect forest fires directly onboard satellites. The system will use a lightweight convolutional neural network to analyze satellite imagery for signs of fire. Rather than transmitting full images for analysis, the proposed system is designed to send essential fire alerts and location coordinates to Earth, reducing detection time, bandwidth use, and energy consumption. Working with community partner SEERAKKU, the project aims to support earlier wildfire detection and faster emergency response while helping protect forests, ecosystems, and communities.
This project was made possible by $1,500 in funding from the Instrumentation and Measurement Society (IMS).
