Using AI to search for unexploded bombs and mines in Cambodia

Post-war consequences are always a complex and costly task for any country. Landmines, explosive materials, located deep, scattered underground are difficult to detect and are a potential threat to the lives of people in the area.

Recently, an AI system has been implemented to assist in searching and remedying unexploded ordnance left over from wartime in some areas of Cambodia adjacent to Vietnam.

The system, developed by computer scientists from Ohio State University, uses object recognition algorithms to detect the specific characteristics of bomb craters, including shape, color, texture and their size. The project is currently undergoing initial testing in Prey Veng province, eastern Cambodia, to find craters in a village in an area that had been bombed by carpet US carpet about 30km from the Vietnamese border.

The actual results show that the AI ​​system has increased the likelihood of detecting craters by 160% more than standard target recognition algorithms, according to a study published in PLOS ONE. The researchers then combined the results with the military records of the area's bombings and summarized them into an extremely detailed area map of unexploded ordnance.

New solutions for mine clearance

In March 1969, when the resistance against the US invaders was raging, the US military began deploying secret bombing raids on the Cambodian border area adjacent to Vietnam. After four years of the campaign ended, an estimated hundreds of thousands of tons of explosives were thrown into these areas, many of which have not exploded and are still scattered in the ground, endangering people's lives. people of both Vietnam and Cambodia.

Using AI to search for unexploded bombs and mines in Cambodia Picture 1

Clearing contaminated areas is a dangerous and costly process, both labor and money. The most commonly used method today is to locate explosives in the ground with metal detectors and radars, then conduct manual excavation very time-consuming and dangerous, while the accuracy is not real. the high. AI models like Ohio University, if widely deployed, will open up new, more effective ways of clearing mines and UXO in explosive-contaminated areas around the world.