Combined Person Classification with Airborne Optical Sectioning

TL;DR

This paper introduces a novel airborne optical sectioning technique combined with deep learning to improve detection of persons under forest canopy, enabling real-time, onboard search-and-rescue operations with higher accuracy.

Contribution

It presents a new method of combining multiple AOS classifications to reduce false positives and enhance true detections in occluded environments.

Findings

Significant reduction in false detections.

Boosted true detection rates under occlusion.

Real-time processing at ground speeds up to 10 m/s.

Abstract

Fully autonomous drones have been demonstrated to find lost or injured persons under strongly occluding forest canopy. Airborne Optical Sectioning (AOS), a novel synthetic aperture imaging technique, together with deep-learning-based classification enables high detection rates under realistic search-and-rescue conditions. We demonstrate that false detections can be significantly suppressed and true detections boosted by combining classifications from multiple AOS rather than single integral images. This improves classification rates especially in the presence of occlusion. To make this possible, we modified the AOS imaging process to support large overlaps between subsequent integrals, enabling real-time and on-board scanning and processing of groundspeeds up to 10 m/s.