Scientists from the United States have presented a new AI-based model that can identify the locations of sunken ships from photographs from a drone. This will reduce the number of costly diving operations.

Independent researchers have partnered with the US Navy to teach an AI-based model to recognize sunken ships on the ocean floor from aerial images of airplanes and ships. The computer model showed an accuracy of 92%. The project was used on the coast of the US mainland and Puerto Rico. It is now ready to be used to locate unknown or uncharted shipwrecks.

The first step in creating the model was teaching the computer what a sunken ship looks like. It was also important to teach the computer to distinguish between sunken ships and the topography of the seabed. To do this, the researchers uploaded data on what the shipwreck looks like.

The National Oceanic and Atmospheric Administration maintains a public database of shipwrecks. They also have a large database of various types of images collected from around the world, including lidar images of the seabed. On their basis, the AI-based model was taught.

Scientists noted that the search for sunken ships is important for understanding history. The model, which automatically maps all sunken ships over a large area, can reduce the time and cost of searching for sunken ships using underwater drones or divers.

The Navy’s Underwater Archeology Division is interested in this work as it can help locate uncharted or unknown shipwrecks. Scientists noted that this is a new method in underwater archeology that can be extended to search for various types of underwater archaeological sites, including buildings, statues, and airplanes.