ISI’s Research & Innovation team was able to automatically identify military vessels at Hamad port, Qatar using deep learning algorithms. “The Kingfisher AI based intelligence system was able to distinguish between civilian and military vessels,” explains Dr. Natalie Fridman, director of ISI’s Research & Innovation department.
The system based on deep learning algorithms that study what a military vessel looks. After the learning stage, the algorithms can automatically detect such ships.
“We teach algorithms to identify the warship class based on deep neural network” Dr. Fridman explains. “Another advantage of the system lies in the fact that from the moment it learns about the vessel class, it also identifies the various derivatives based on it.”
frigate vs. destroyer
One of the challenges of intelligence systems for the automatic analysis of satellite images lies in the ability to differentiate between sub-types of battleships. Destroyer, corvette, frigate, research ships, etc.
“At this stage, a human analyst is still required to improve operational analysis,” explains Dr. Fridman. “Over time, our ambition is to improve the algorithms so that they can automatically identify military vessels with a few errors.”
Situational awareness in the marine domain
The ISI’s KingFisher system used as a support system for analysts. The ability to distinguish between civilian and military tools makes it possible to scan large areas of the sea by satellite and to quickly draw conclusions from the photographs.
The system enables rapid scanning of seaports, military or civilian, to obtain an up-to-date deployment picture of enemy naval power in a defined area.
Additional use of the system is to gain situational awareness in the exclusive economic zone (EEZ). Rapid scanning of the marine arena enables monitoring of threats to energy resources, piracy, illegal fishing, enemy ships or any element that threatens to violate state sovereignty.
Lobby image: pixabay
— imagesatintl (@imagesatint) March 20, 2018