A trio of researchers from Johannes Kepler University has used artificial intelligence to improve people’s search for lost in the forest using thermal imagers. In an article for the journal Nature Machine Intelligence, scientists share how they applied the deep learning network and how well it works.
When people are lost in the forests, search and rescue specialists use drones to survey areas where the missing maybe. Rescuers also use binoculars and thermal imagers. Unfortunately, in some cases, thermal imaging does not work properly due to vegetation covering the soil. The heating of trees from the Sun to a temperature close to a lost person’s body temperature also interferes. In the new work, the researchers sought to overcome these challenges by using deep learning to enhance thermal imagers’ images.
The new system uses an AI application to process multiple images of a specific area. Comparison and processing of data from different cameras allow several thermal imagers to work as one large telescope. After processing the AI images, the final terrain images have a higher depth of field. But the footage shows how the tops of the trees seemed blurry, and the outlines of people on the ground were more recognizable. To train the AI system, the researchers had to create their own image database. They used drones to photograph volunteers on the ground in a variety of positions.
System testing has shown that its accuracy is 95% compared to 25% of traditional thermal imaging images. The system is ready for use by search and rescue teams and can also be used by law enforcement, military, or wildlife conservation groups, the scientists conclude.