Driven by AI and CV, The way of urban management has been widely changed. How to efficiently carry out 3D digitization and how to connect 3D world and digital information sources has become a key challenge restricting AI city management. Based on years of scientific research, combined with our core algorithms annd software-hardware integration capabilities, Stereye developed modeling, connecting, and analyzing technology that targets the in-depth application dilemma of future digital cities.
3D can effectively enhance the ability of video intelligent vision. 3D information can effectively correlated video data in different points. As a result, local governance and management can be improved to global.
In the lifetime of a building, the digital twins involving floorplans/BIM will greatly expand its value. Through the autonomous floorplan/BIM and 3D data registration and calibration algorithm, AI can produce construction progress, installation error, layout detection and other inspection analysis ability.
We firmly believe that the recognition algorithm of a single data source will eventually be replaced by the multi-source AI algorithm. The joint processing capabilities of multi-source data (including 3D maps, 3D/2D videos, non-traditional optical sensors and IoTs, etc.) will promote urban building management to a higher dimension.
In order to realize the AI system of multi-source data, the 3D system will become the central node for linking different data. The problem brought by this is how to add 3D positioning and calibration information to non-3D data. Traditional correlation algorithms mostly rely on complex manual operations and more 3D vision technology is required to provide better solutions.
This algorithm realizes automatic recognition and matching of floorplans/BIM and scanned 3D model. With buildinng semantic recognition established by the neural network, AI can localize those data under minimized manual intervention.
This system can optimize the positioning of video surveillance in a known 3D environment. This solution does not rely on manual intervention, and can achieve centimeter-level video surveillance localization in a static environment in typical scenarios.
This system provides precise positioning in real time while scanning the environment. Under this framework, third-party systems such as QR code scanners and asset inspection sensers can be connected to the 3D system. This solution does not rely on an external positioning device, and maintains stable operation for closed scenes such as indoor, underground, and tunnels.
For consumer-level indoor localization scenes, it provides AR localization and navigation capabilities using 3D semantic vision and tracking technology. This solution is compatible with common mobile device platforms, achieving high-accurate navigation for indoor and underground scenarios.