Researchers from Johannes Kepler University in Austria are using artificial intelligence (AI) to improve drone searches for people lost in the woods.

To enhance the capabilities of thermal imaging cameras — used in most overhead search and rescue operations wherein infrared radiation emitted from an object is captured by the camera — the team used a deep learning application to highlight the body temperature differences between volunteers in the woods and their surroundings as captured by drones flying above.

Commonly, thermal imaging cameras are employed in overhead searches for those missing in the woods, yet they are not always effective for capturing temperature differences between trees and people due to the potential for the sun heating treetops or vegetation on top of soil to temperatures similar to those of the human body, thus complicating search results.

To prevent this confusion, the team captured several drone images of an area under observation where volunteers were positioned and used an AI application to create as series of images that gave the appearance of an image captured by a significantly larger lens.

Once processed, those images offered a higher depth of field wherein the treetops were blurred but volunteers on the ground were significantly easier to spot.

During tests of the system, the team determined that the AI enhanced thermal imaging cameras were 87% to 95% accurate whereas standard thermal imaging alone is roughly 25% accurate.

According to the researchers, this technology is ready for use and appropriate for law enforcement, military, search and rescue, and wildlife management applications.

The technology is detailed in the journal Nature Machine Intelligence.

For more information, watch the accompanying video that appears courtesy of Johannes Kepler University.

To contact the author of this article, email