Deep Mirror develops spatial intelligence technology that creates multi-layer digital twins of the real world through high-precision 3D mapping and real-time semantic segmentation. Their platforms, including MirrorScene and MirrorSpace, enable collaborative augmented reality experiences and city-scale spatial localization, addressing the need for accurate and interactive virtual environments in various applications.
Funding
Funding not disclosed
Founders
Product
Problem
Creating accurate and interactive virtual environments for applications like augmented reality gaming, city-scale navigation, and collaborative design requires precise 3D mapping and real-time spatial understanding. Existing solutions often lack the precision, scalability, and real-time capabilities needed for seamless integration and user experience in diverse environments.
Solution
Deep Mirror offers a suite of spatial intelligence platforms that create multi-layered digital twins of the real world. MirrorScene enables collaborative augmented reality experiences through real-time 3D scanning and semantic segmentation. MirrorSpace provides city-scale spatial mapping and localization using various sensor inputs and a self-learning algorithm. MirrorSuite delivers on-device simultaneous localization and mapping (SLAM) solutions with high precision and cross-platform compatibility. These platforms address the need for accurate, interactive, and scalable virtual environments across various mobile and AR/VR devices.
Target Audience
The primary audience includes developers and enterprises in augmented reality gaming, navigation, advertising, and other spatial computing applications requiring precise and scalable 3D mapping and localization solutions.
Features
- **MirrorScene:** Collaborative 3D scanning tool supporting multi-user spatial mapping, real-time 3D visual reconstruction, and shared spatial anchor creation.
- **MirrorSpace:** City-scale spatial mapping platform using Lidar, 360-degree cameras, and mobile device cameras with a cloud-based storage and computing system.
- **MirrorSpace:** Self-learning algorithm adapts to both high-quality and user-generated data, processing multi-gigabytes of data daily and handling hundreds of requests per second.
- **MirrorSuite:** On-device SLAM solution with less than 1% relative precision and absolute precision under 2 cm and 1 degree.
- **MirrorSuite:** Supports various sensor setups (IMU + mono/multi cameras) and offers comprehensive calibration solutions for AR/VR devices.
- **MirrorSuite:** Cross-platform compatibility across hardware platforms like RK and Qualcomm, leveraging DSP/GPU/NPU acceleration.