In order to prevent accidents at railroad crossings, Seibu Railway will introduce a new system next week that detects people left behind by AI and 3D image analysis.
This is the first time in Japan that a 3D camera-based railroad crossing anomaly detection system has been put to practical use.
There were 1,071 accidents at railroad crossings in the five years up to last year, and 454 people were killed.
Seibu Railway, which has been developing with electronics manufacturers, has newly developed AI to detect human skeletons from images taken by cameras at railroad crossings, and 3D camera image analysis of railroad crossings to recognize people based on their height and thickness. However, we decided to implement two systems.
In either case, if a person or bicycle remains inside the railroad crossing for a few seconds after the crossing gate begins to move, an abnormality is detected and a stop signal is issued to the train driver to notify the danger.
Seibu Railway will start operation from the 14th of this month at three small railroad crossings where cars do not pass.
According to the Ministry of Land, Infrastructure, Transport and Tourism, the practical application of a system that specifically detects people at railroad crossings is part of AI, but it is the first in the country to use a 3D camera.
Takaharu Inuzuka, head of Seibu Railway's signaling and communication division, said, "Until now, we had to ask someone bystander to press the emergency button, but by detecting at night and when there are no people, we can reduce accidents at railroad crossings even a little. I want to keep going."