AI developed to automatically detect seabed plastic waste

A group of researchers in Japan has developed an artificial intelligence system that automatically detects plastic and other waste at the bottom of the ocean in the hopes of helping environmental cleanup efforts.

The group at the Japan Agency for Marine-Earth Science and Technology, or JAMSTEC, developed the system by utilizing deep-sea videos that it had taken in the past. The researchers surveyed seabeds around Japan for more than 40 years with unmanned probes and other devices. They extracted more than 12,000 images from the videos, analyzed the shape and other information about plastic waste and fed the data to a computer to learn the characteristics.

The researchers say the system can recognize and count visible pieces of waste from 2 centimeters to 1 meter in size, but cannot recognize smaller ones, including microplastics.

The group says measures are necessary to tackle the problem of plastic waste on the seabed because they remain in the environment without degrading. Researchers previously had to visually detect the location and the amount of the garbage by replaying and pausing videos repeatedly. They say the new AI system has reduced the necessary time by more than half.

Nakajima Ryota, a senior researcher at JAMSTEC, said the accumulation of nondegradable plastic garbage poses a serious problem that will worsen environmental contamination, unless measures are taken. He added that his group plans to improve the accuracy of the AI system and utilize it to quickly take action.