Sponsor(s)
National Science Foundation
Contributor(s)
Sanjit A. Seshia
Remote Video
Submitted by Sanjit Seshia on May 14th, 2020
Submitted by Sanjit Seshia on January 10th, 2020
Explanation of Demonstration: Scenic is a new domain-specific probabilistic programming language for specifying the environment of a cyber-physical system. Scenic can be used to describe scenarios of interest and automatically generate realistic data from them for testing or training purposes. We will demonstrate how to use Scenic to generate images of traffic scenarios to improve the performance of an object-recognition neural network for autonomous driving.
Submitted by Anonymous on January 22nd, 2019
Submitted by Sanjit Seshia on December 4th, 2018