Core Technologies

Based on the foundation of proprietary technologies and a core “brain” built on a deep learning platform, SenseTime has rapidly opened up AI application in multiple vertical scenarios.

SLAM and 3D Vision

SLAM and 3D Vision

Technical Capabilities

Flat/3D Object Recognition and Tracking
Simultaneous Localization and Mapping (SLAM)
Structure from Motion (SFM)
Real-Time Dense 3D Reconstruction
Lighting Estimation
Lightweight Cross-Platform AR/VR Engine

Flat/3D Object Recognition and Tracking

The advanced flat/3D object recognition and tracking system quickly identifies flat and 3D objects to provide stable real-time object tracking.

Simultaneous Localization and Mapping (SLAM)

The self-developed, industry-leading visual inertial navigation SLAM system offers real-time 3D object pose estimation and real-time object tracking while simultaneously constructing a 3D geometric map of the scene online. SLAM supports real-time positioning and virtual object integration in real scenes on mobile platforms, achieving a perfect integration of virtuality and reality.

Structure from Motion (SFM)

The structure from motion (SFM) system is capable of conducting camera motion recovery and 3D object reconstruction for a sequence of more than 100,000 images and videos, making it ideal for photo tourism and 3D roaming deployment.

Real-Time Dense 3D Reconstruction

Completing dense 3D reconstruction of natural scenes online, the solution supports real-time virtual 3D scene digitization, realistic rendering, shadow carving, and other integration effects on mobile devices, including sheltering and collision effects, perfectly realizing virtual reality.

Lighting Estimation

The lighting estimation solution predicts the lighting condition of the scene for realistic rendering of the virtual objects.

Lightweight Cross-Platform AR/VR Engine

The lightweight cross-platform AR/VR engine supports multiple platforms, lighting and material models, virtual object sheltering and shadow projection to enable various expression-driven facial animations of avatars.