By means of synthetic data generation, large amounts of image- and metadata can be extracted directly from a virtual scene, which in turn can be customized to meet the specific needs of the algorithm or the use-case. Furthermore, the use of virtual objects avoids problems that might arise due to data protec- tion issues and does not require the use of expensive sensors. We propose a framework for synthetic test data generation utilizing the Unreal Engine.
Unreal Engine
The Unreal Engine provides a simulation environment that allows one to simulate complex situations in a virtual world, such as data acquisition with UAVs or autonomous diving.
Computer Vision
Built-In Generators allow a large vareity of images to be synthesized. In addtion to basic color images, depth images, semantic segmenations, normal and depth maps. The Image Generators can also be used in conjungtion to create multi camera setups.
Deep Learning
Generate accurate grountruth, for machine learning tasks! Bounding boxes, pose data(WIP) and various other properties of scene objects can be extracted
Fast Setup & Configuration
By utilizing the UI of the Unreal Editor complex data generation constructs can be created in minutes, without touching a single line of code.