Unmanned Aerial Vehicle for Search, Rescue, and Evacuation Using Deep CNN Detection

Clients

Techstack

Purpose

When someone goes missing after a disaster, searching on foot is slow, dangerous, and every hour counts. A research team wanted to prove that a small drone could take over the first sweep — spotting people and animals from the air on its own, so rescuers know where to go before they ever set foot on the ground.

Description

We built a system that flies a compact drone, reads its live camera feed, and uses a deep-learning vision model to pick out humans and animals in real time, marking each find on screen and logging when and where it was seen. A pilot watches the detections as they happen and can steer the search from a simple control interface. It is exactly the kind of hard, safety-critical computer-vision problem — reliable detection from a moving aerial camera — that we take on and ship.