VIVID - Virutal Environment for Visual Deep Learning


VIVID (VIrtual environment for VIsual Deep learning) is a photo-realistic simulator that aims to facilitate deep learning for computer vision. We have won the 2018 ACM MM Best Open Source Software Award (github). In the past few years, we keep developing new applications and have two main projects now: Pothole Game and Drone Battle: Drones vs Humans. Our projects are based on Unreal Engine and Microsoft AirSim. Students need to be familiar with C/C++, Unreal, and learn to compile AirSim code first.

Learning Steps
  1. Unreal Tutorial
  2. Microsoft AirSim
Pothole Game
Paper Tsai et al., Learning Pothole Detection in Virtual Environment, CACS, 2020
Code pothole game v0.3 (UE4.25)
Drone Battle: Drones vs. Humans Code drone_battle_v0.1 (UE 4.25)