AWL, Inc.
Final Report
AWL, Inc. project conducted from February 2025 at the Sumida City Urban Development Department, Parks Section / Sumida Skateboard Park was completed in March 2025.
AWL, Inc. implemented a project at the Sumida Skateboard Park using the company’s edge AI video analytics technology for problem behavior detection to encourage helmet wearing among park users and maintain a safe environment.During the project period, three edge AI cameras and two speakers were installed at the park’s entrance and play area.
When a “non-helmeted user” stayed within the camera’s detection area beyond a certain period, the system played automated audio alerts via speakers installed next to each camera or live voice alerts from city staff located remotely. This enabled detection and prevention of problem behaviors without requiring city staff to visit the site.
Results: For detection accuracy by camera, the entrance camera recorded 86.2% correct detection for non-users and 63.6% for users (over four days), and the play area camera recorded 93.0% correct detection for non-users (over seven days). Factors contributing to false detections included confusion between hats and helmets and daytime sunlight effects.
Experiment Results for the Entrance Camera
Experiment Results for the Play Area Cameras
We also confirmed the effectiveness of the automated audio alert and staff alert (remote live voice alerts) system by checking the system operation rate and conducting a survey.
The alert system had an operation rate of 100% during the project period.
Operation Log Totals
Survey questions for city staff involved in the project included:
- Did the AI system for helmet detection provide material for considering future workflow improvements?
- How did staff workload change after the introduction of the AI system?
- How effective did you find the alerts by staff located remotely?
- How effective did you find the AI system’s automated audio alerts?
- What is the overall assessment of this initiative?
Survey results indicated that using AI technology yielded insights on improving park entry management methods, city staff monitoring and management tasks, and announcement content/methods leading to behavioral change among users, including non-helmeted users.
Results of the Survey
Going forward, we will improve problem behavior detection technology based on results from this project and promote introduction through public procurement.
