Skip to main content

Event Details

  • Monday, November 6, 2017
  • 11:00 - 11:30

UE4Sim: A Photo-Realistic Simulator for Computer Vision Applications

In this talk, I will present UE4Sim, a photo-realistic simulator based on the open source computer game engine, Unreal Engine (UE4). UE4Sim was developed by the Image and Video Understanding Lab (IVUL) at KAUST and it is designed to be a fully enclosed environment to train, validate, and evaluate methods developed for a multitude of computer vision, machine learning, and robotics tasks, including autonomous navigation, visual object tracking, action recognition, crowd understanding, to name a few. One important feature of UE4Sim is that it provides a desirable level of photo-realism, which can ultimately facilitate the transfer of models/knowledge acquired in the simulator to the real-world, thus, avoiding the need to retrain from scratch in the latter environment. I will highlight the merits of UE4Sim by focusing on three exciting applications: persistent aerial tracking, self-driving cars, and autonomous racing UAVs (unmanned aerial vehicles).