JAPAN Artificial intelligence against bears in Japanese cities

Sightings and attacks on people are increasing: this year 56 incidents and two deaths have already been recorded. To control the phenomenon, several municipalities are adopting systems based on video cameras “trained” to automatically identify if the moving figure is a bear, and raise the alarm.

Tokyo (/Agencies) – In Japan, the number of bears entering urban areas in search of food is also growing considerably. The agency Kyodo reports that in an attempt to reduce the danger posed to humans, several Japanese municipalities have begun using an artificial intelligence-assisted surveillance system. The cameras were installed along river banks, on mountain roads and in residential areas with the collaboration of private companies. The images captured by these devices are analyzed 24 hours a day to notify authorities and residents of bear sightings, helping to ensure public safety.

Preliminary data from the Ministry of the Environment speaks of 10,704 bear sightings between April and July, far exceeding the 8,536 cases from the same period last year, which already saw the highest number of human victims ever recorded in Japan. In July, 2,471 bears had been killed or captured, and between April and August, 56 incidents with human casualties occurred, affecting 58 people, including two deaths.

The brown bears of Hokkaido and the black bears of Honshu and Shikoku are the two species that populate Japan and live in at least 34 prefectures, according to the ministry. In recent months, national and local headlines have reported incidents of people encountering bears, and one of them was seen playing in a river in a popular tourist town.

The surveillance system works based on cameras that detect moving objects. The data is then sent to artificial intelligence, which has been trained with around 50,000 images of bears and sends an alarm only if it identifies the moving figure as such.

The adoption of this system in Komatsu City has apparently reduced the heavy reliance on local hunters, who are traditionally the main resource for dealing with threats from bears and other animals. The Environment Ministry also began testing in Toyama Prefecture to determine whether the AI-based pointing system could detect bears using images from existing surveillance cameras.

Hiroshi Saito, a professor at Aizu University in Fukushima Prefecture, points out that bears tend to frequent residential areas, regardless of how many times they are chased away. However, Saito is not opposed to a more aggressive approach, and has developed a system that emits powerful lights and sirens to scare away the animal when detected by a camera controlled by artificial intelligence. Twenty such cameras were installed in six locations in Fukushima Prefecture to determine the effectiveness of the system.