What’s ANYVERSE
Data-as-a-Service
Software-as-a-Service
The ANYVERSE synthetic data solutions offer the speed, scalability, and visual fidelity needed by machine-learning teams to rapidly progress to high-confidence perception models.
By using a proprietary physics-based render engine, ANYVERSE simulates the visual appearance of the real-world with higher accuracy and variability than is practical with other approaches. Balanced, rich data that mirrors sensor capabilities and provides photometric quality.
ANYVERSE is not a dataset, it is the dataset you need.
Applications
Autonomous Vehicles
Drive-Assist (ADAS)
Autonomous Robotics & Smart Technologies
Our difference
High Fidelity
Game engines do not accurately mirror reality. ANYVERSE’s physics-based spectral unbiased renderer provides accurate visual quality and ground-truth representation of lighting and the environment.
Faster Cycles
Speeding up AI training-testing cycles is critical. With ANYVERSE faster perception training-testing iterations are possible, guaranteeing a competitive advantage over teams relying exclusively on real-world datasets.
High Control
Fine-grained control of simulation features and variations is key to produce rich datasets in an agile way, increasing confidence and robustness in the perception system.
Flexible Solution
The best part – there is no need to master complex software. Clients can team up with our engineers to produce custom datasets or get access to a highly scalable SaaS platform to produce the datasets they need.
Core technology
Predefined and procedurally-generated 3D scenes for urban and suburban areas.
Large library of 3D assets: vehicles, pedestrians, buildings, traffic lights, signs, vegetation, etc.
Agent-based models for traffic and pedestrians.
Proprietary unbiased spectral render engine for accurate optical and light simulation.
Physically accurate camera model and support for multiple cameras.
Physically accurate atmospheric and weather models.
LiDAR model based on real-world devices.
Automatic generation of annotated data and pixel-accurate segmentation for every element.
2D/3D bounding boxes.
Multiple data channels: RAW XYZ32 (sensor buffer), RGB32, color screen, object ID, material ID, depth, normal, 3D positions, etc.