Launch of Field Trial for Bear Detection System Using AI-Equipped Cameras at Base Stations

NTT Docomo has initiated a field trial on May 22, 2026, using surveillance cameras integrated with image recognition AI at base stations in Hokkaido to detect bear sightings in real-time. This project leverages Docomo's communication infrastructure and AI technologies to balance nature conservation with human safety, aiming to contribute to a 'Nature Positive' future. It is part of the 'Medium-Term Roadmap for Biodiversity' formulated in May 2025.

■ Background
Due to the recent expansion of bear habitats, bear-related personal injury incidents exceeded 230 cases in fiscal 2025, with 13 fatalities. In fiscal 2026, following hibernation, 7 emergency hunting-related discharge incidents have already occurred, indicating an intensifying situation. Furthermore, challenges such as the aging and shortage of hunters, who are key to countermeasures, have become evident.
Given the limitations of manual monitoring in vast regions, transitioning to a comprehensive defensive system—incorporating ICT-based monitoring and collaboration with relevant agencies—is urgent to ensure resident safety.

■ Project Overview
Surveillance cameras integrated with image recognition AI will be installed at two base stations in Hokkaido to detect bear sightings by analyzing footage in real-time.

Feature 1: Utilization of Base Station Facilities
Stable mobile networks are established around base stations, allowing for real-time video transmission and rapid notification even in rural areas. Leveraging existing facilities minimizes environmental impact and deployment costs, offering the advantage of future large-area coverage.

Feature 2: Real-time Detection via Image Recognition AI
Footage from cameras at base stations is analyzed by image recognition AI to detect bears. To ensure 'urban bears' are not missed, the system aims for stable detection performance under varying environmental conditions, including lighting changes. High-speed inference AI is operated on low-latency docomo MEC to enable real-time detection.

■ Trial Period
Scheduled from May 22, 2026, to November 30, 2026.

■ Future Outlook
Through the trial in Hokkaido, Docomo will verify the system's effectiveness and accuracy, aiming for future deployment to municipalities struggling with bear damage. Docomo plans to implement a comprehensive system capable of mapping bear positions, notifying relevant agencies promptly, and issuing instructions for deterrent sounds. This initiative aims to contribute to a safer, more comfortable society by mitigating conflicts between wildlife and humans.

Additionally, Docomo is developing new electric fence systems and related technologies to prevent wildlife from encroaching on agricultural and residential areas. By maintaining appropriate distances from wildlife, including bears, Docomo contributes to the realization of a safe society.