Synthverse | Synthetic data simulation platform

Synthverse is a synthetic data platform that can meet the synthetic data requirements of various usage scenarios and AI training for different corner cases. Synthverse can automatically generate 3D scenes based on high-precision maps, restore specific scenes with high fidelity through 3D reconstruction technology, and can generalize the dynamic and static elements of the 3D reconstructed scenes through the generalization tool to improve the scene coverage. It can also create diverse and complex scenarios through generative AI.

High Fidelity 3D Environment toolchain + 3D Asset Library

Traffic signs, road facilities 6000+Traffic signs, traffic lights, road markings, sidewalks, fences Vehicles 200+ vehicles Trucks, commercial vehicles, family sedans, family SUVs, special vehicles, tricycles, bicycles Vegetation 1000+ trees, 30+ types of vegetation SOTIF road falling/accumulation/abnormal vehicle painting/status/pedestrian clothing/movement

Automatically generates
Perceptual testing simulation scenes

Use procedure generation techniques Based on HD map Supports roads and surrounding elements Can be configured to different styles

Physical-level Sensor models

Laboratory camera parameter calibration Parameter simulation (distortion, fisheye, color restoration, dynamic range, white balance, glare, color difference, etc.) Scene factors (based on physical lighting models, complex weather systems) Abnormal situation (obstruction, dust, liquid splashing)

Synthetic Data Platform - Synthverse

High Fidelity 3D Environment toolchain + 3D Asset Library

Automatically generates Perceptual testing simulation scenes

Physical-level Sensor models

Automatically generates
Perceptual testing simulation scenes