Twitter
Google plus
Facebook
Vimeo
Pinterest

Fluid Edge Themes

AI Improves Search And Rescue Accuracy

Home  /  News   /  AI Improves Search And Rescue Accuracy
AI Assisted Search And Rescue

AI Improves Search And Rescue Accuracy

AI Assisted Search And Rescue

AI Improves Search And Rescue Accuracy

Researchers at the Johannes Kepler University in Linz have developed a new AI model that combines aerial recordings from multiple drones/planes/helicopters and processes them to improve the accuracy of human recognition in heavily forested areas, helping find people lost in the woods.

Their model made use of a deep learning network trained by real-world images of people hiding in wooded areas. The artificial intelligence combines multiple regular RGB and thermal images to create a depth-enhanced final image that removes focus from treetops and places it on the ground, where lost people tend to be located. It then analyzes this composite image to find the hidden human forms. 

While thermal imaging technology typically does a good job separating warm humans from their surroundings, there are times when tree canopies can obscure a person’s thermal signature. At other times, the surrounding flora can be of similar temperature as a human, meaning they do not stand out from their surroundings when using a thermal camera. 

According to the creators, David Schedl, Indrajit Kurmi and Oliver Bimber, the system is up to 95% accurate, compared to 25% accuracy when using thermal imaging cameras alone.