Synthetic 3D Video Generation for Machine Learning | 2017-2018
In this joint project with Korea National University of Transportation, we developed a 3D computer graphics based driving scene simulator that researchers can use to test computer vision algorithms used in self-driving cars. The simulator can generate photorealistic 3D driving scenes of a variety of conditions with user inputs in real-time so that the researchers can use the generated video clips for training and testing their computer vision model.
Tools: Unity 3D, Maya
Hyung Gyu Oh, Jihyun Park, Sieun Park, Dayoung Lee
Ho Gi Jung, Jae Kyu Suhr, Kyongtaek Choi, Hyung Gyu Oh, and Jusub Kim, HandSpan-EX: Sensor Fusion-based Precise Localization System for Highway Driving, The Korea Society of Automotive Engineers Fall Conference, 2017