Sonid announced on the 5th that its subsidiary, Sonid AI, has completed the development of its 'AI CCTV solution,' which integrates its visual language model (VLM) technology and blockchain security technology into the on-device artificial intelligence (AI) 'BrainBot.'
BrainBot has improved the speed of CCTV data processing without relying on a central server through on-device AI technology. It maintains stable performance even in the event of network disruptions. The company explained that, in particular, the application of its own VLM technology and blockchain security technology has enhanced object and situation recognition abilities and strengthened personal data protection for data stored in the cloud during central control.
According to the company, the AI CCTV solution was developed to database human movements, making it easier to analyze the causes of accidents and incidents. To achieve this, VLM technology was applied to learn from video and images and generate results for application. When VLM technology is applied, CCTV can recognize objects in real time, detect abnormal behavior, and perform more sophisticated analyses.
Sonid AI has developed its own multimodal AI model VLM that can understand and search videos. It can identify people and objects in the video and determine their actions. VLM is an enhanced large language model (LLM) with image and video processing capabilities. For example, if you search for 'find a person holding a weapon' or 'find a person jaywalking at the crosswalk' among the footage captured by CCTV, it will find the desired scene and summarize the content in text.
A Sonid AI official said, 'We expect it to be widely adopted not only for government and local governments' crime prevention and security CCTV, police vehicle searches, smart city infrastructure projects, and private security corporations but also to replace existing CCTV systems in locations where crime prevention and real-time monitoring are necessary.'