Simulation and Synthetic Data Generation

Simulation and synthetic data generation (SDG) are increasingly central techniques to augment the dataset of perception models and to provide a digital twin for end-to-end development & validation of applications.

The goal of SDG is to generate a flexible number of well-placed cameras that provide full scene coverage, along with corresponding ground truth data, to support Multi-Camera Multi-Target Tracking (MTMC) use cases. Detailed dataset introduction can be found in our released PhysicalAI-SmartSpaces.

Briefly, you would need to do the following steps to make use of synthetic data in MTMC use cases.

• Development Tools Setup
• Synthetic Data Generation
• Scene Customization
• Dataset Post-Processing
• Sim2Real Data Transfer

As part of the standalone deployment mentioned, please refer to the Download Warehouse Artifacts section in Quickstart Guide for instructions on downloading the standalone deployment.