AI discovers a shipwreck from the surface of the sea

The· Research briefs A short view of interesting academic research.

Big idea

Collaboration with the US Navy Underwater Archeology Branch, I taught my computer how to recognize a shipwreck on the ocean floor from scans made by aircraft and ships on the surface.The· Computer model we created 92% accurate to find known shipwrecks. The project focused on the continental United States and the coast of Puerto Rico. Now you are ready to use it to find unknown and unmapped shipwrecks.

The first step in creating a wreck model was to teach the computer what the wreck would look like. It was also important to teach the computer how to tell the difference between the debris and the topography of the seafloor. To do this, we needed a lot of wreck examples. We also needed to teach the model what the natural ocean floor would look like.

Conveniently, the US National Oceanic and Atmospheric Administration Public database of shipwrecks.. There is also a large public database of different types of images collected from around the world, such as sonar and sonar. LIDAR Image of the seabed. The image I used extends a little over 14 miles (23 km) from the coast and extends to a depth of 279 feet (85 m). This image contains a vast area without shipwrecks and occasional shipwrecks.

Important reason

Finding a wreck is important for understanding the human past-think trade, migration, war-but Underwater archeology is expensive and dangerous.. A model that automatically maps all shipwrecks over a large area can reduce the time and cost required to find a shipwreck, either underwater drone or a human diver.

The Navy’s Underwater Archeology Department is interested in this task as it may help find unmapped or unknown Navy shipwrecks. In a broader sense, this is a new method in the field of underwater archeology that can be extended to look for different types of underwater archeology features such as buildings, statues and planes.

What other research is being done in this area?

This project is the first archaeologically focused model built to automatically identify shipwrecks on a wide area, in this case the entire coast of the continental United States. There are several related projects focused on the discovery of shipwrecks. Uses deep learning and images collected by underwater drones.. These projects can find a handful of shipwrecks in the area just around the drone.

What’s next

We would like to include more wreck and image data from around the world in our model. This allows the model to recognize different types of shipwrecks very well. We also hope that the Navy’s Underwater Archeology Department will dive into some of the places where the model detected the wreck. This allows you to check the accuracy of your model more carefully.

I’m also working on several other archaeological machine learning projects, all of which are built together. The overall goal of my work is to build a customizable archaeological machine learning model. This model allows you to quickly and easily switch between predictions of different types of archaeological features on land and underwater in different parts of the world. To this end, I am also working on a project focused on finding ancient Mayan archaeological structures, Mayan caves, and Romanian burial mounds.

Author: Leila Character-University of Texas at Austin, Liberal Arts, Ph.D. in Geography

AI discovers a shipwreck from the surface of the sea

Source link AI discovers a shipwreck from the surface of the sea

Related Articles

Back to top button