Talk Proposal: Towards the Realistic Evaluation of Evasion Attacks using CARLA

Abstract

In this talk we describe our content-preserving attack on object detectors, ShapeShifter, and demonstrate how to evaluate this threat in realistic scenarios. We describe how we use CARLA, a realistic urban driving simulator, to create these scenarios, and how we use ShapeShifter to generate content-preserving attacks against those scenarios.

Submitted as talk proposal to Dependable and Secure Machine Learning (DSML '19)

Cited by in corpus (1)

• DriveGuard: Robustification of Automated Driving Systems with Deep Spatio-Temporal Convolutional Autoencoder