Fig. 1
From: Digital reality: a model-based approach to supervised learning from synthetic data
The Digital Reality Concept. The process starts by creating partial models of reality by modeling, capturing, or learning individual aspects such as geometry, physical properties including materials, behavior, or lighting. The partial models are composed of parametric scenarios by manual configuration, data fitting, or machine learning. The composition is performed in such a way that all aspects of reality relevant to a specific simulation are provided. Setting all parameters of such a scenario to fixed values creates a concrete instance of the scenario, corresponding to a simulation-ready 3D scene. The scene is then rendered by a forward-simulation of the imaging process. The resulting synthetic images are used to train a machine learning system