Restroomguard Savvy: 프라이버시 우선 커뮤니티 보안 AI

In sensitive spaces such as public restrooms and locker rooms, video surveillance like CCTV has been limited due to concerns about privacy infringement. As a result, even if violence, falls, or spycam crimes occur in restrooms, it is difficult to identify them in real-time, and there are limitations in responding quickly in emergencies.

In the past, only emergency bells or human detection sensors were used, but they could not detect intentional crimes (assault, spycam installation, etc.) or screams, and it was difficult to check for abnormal situations such as prolonged occupancy or smoke generation. In addition, even if sensors were installed, vulnerable groups (women, the disabled, the elderly, etc.) were still exposed to anxiety due to malfunctions (e.g., misinterpreting the sound of urinals as screams) or limitations in preventing crime.

In short, the need for a new approach that can ensure safety while protecting privacy has emerged.

The biggest differentiator of RestroomGuard Savvy is that it detects various dangerous situations with high-precision AI without using any cameras. An AIoT terminal that combines a Time-of-Flight 3D depth sensor, sound analysis (scream and abnormal noise detection), and air quality sensor monitors the state of the space 24 hours a day.

For example, it detects falls (a person does not move for a long time), assault patterns (sudden violent movements and screams), illegal filming attempts (strange device movements), and smoking/fire smoke in real-time. Upon detection, it immediately transmits only anonymized event signals to the management system and emergency center (PSAP). Original video and audio are never stored or transmitted, and only AI-interpretedmetadata is encrypted and transmitted (mTLS)to thoroughly protect privacy.

Unlike competing solutions, this vision-free AI safety management achieves very high accuracy without the concern of privacy infringement (field test results show 93.24% precision and 95.10% recall). It also features disabled emergency notifications (visual, auditory, and vibration) and remote OTA updates for operational convenience.

In summary,"Safety that protects without seeing"is implemented, and the key is that it captures both privacy and safety compared to similar products.

The main adopters are expected to be public and commercial facility managers and local governments in the B2G/B2B area. For example, thepublic restroom management entities (local governments, public institutions)or companies operating multi-use facilities such as department stores and hospitals are likely to lead the installation.

They have the responsibility for the safety of actual users and are legally liable in the event of an accident, so they have an incentive to invest in safety even if it means paying for this system. It can also be applied to elderly care facilities or facilities for the disabled, so related public welfare fields (B2G) may also be interested.

Rather than individual consumers purchasing it, it is expected to be introduced in a batch at the facility level (Classification: B2B/Public - Local governments and facility operators). Currently, 127 units have already been piloted in 23 public restrooms in Korea and abroad to verify the effect, and there have been export cases to Japan, Malaysia, and Switzerland, so the adopting entities are expanding to global smart city-related organizations.

This solution can be applied to various spaces beyond restrooms. It is easy to introduce even in countries with strict legal regulations or privacy-sensitive environments because it does not use cameras.

It has already been exported to overseas countries such as Japan and Malaysia and has undergone local demonstration. In the future, it plans to expand its market to Singapore, Europe, and North America. Technically, it can be expanded to safety monitoring of hospital rooms and elderly living alone, or worker protection in industrial sites by changing the AI model.

In addition, uniuni announced plans to move beyond the hardware combination model to an AI software (SaaS)-centric structure and apply it to various sensors/platforms.

However, localization is required for emergency response system linkage and local language voice recognition in each country. In some areas, even acoustic sensors may be sensitive due to privacy concerns, but this can be addressed with anonymization design.

Overall, it is a technology with the potential to continuously expand its application fields and regions in line with the smart city and public safety trends.

At CES 2026, this product was selected for its innovation in Human Security for All and smart community fields. The judges highly praised the fact that"safety can be maintained without infringing on privacy,"and mentioned it as an exemplary case of using AI technology to protect human rights.

Foreign media such as Fast Company also paid attention to the emergence of AI safety management without cameras, and evaluated the achievement of public safety without personal information as an innovation. In terms of technical completeness, it has been recognized as a reliable solution with actual field application figures and verification in many countries.

Market expectations are also quite high in connection with smart city and safety budgets in each country. However, some point out the adoption speed of the conservative industry (facility management field) or the management of potential false positives, and there is a cautious view that it may take time for actual commercial spread.

Nevertheless, the overall evaluation is summarized as"Harmony of Privacy-Tech and Safety-Tech"with favorable reviews.