DriveGuard: Robustification of Automated Driving Systems with Deep Spatio-Temporal Convolutional Autoencoder
arXiv:2111.03480 · doi:10.1109/WACVW52041.2021.00016
Abstract
Autonomous vehicles increasingly rely on cameras to provide the input for perception and scene understanding and the ability of these models to classify their environment and objects, under adverse conditions and image noise is crucial. When the input is, either unintentionally or through targeted attacks, deteriorated, the reliability of autonomous vehicle is compromised. In order to mitigate such phenomena, we propose DriveGuard, a lightweight spatio-temporal autoencoder, as a solution to robustify the image segmentation process for autonomous vehicles. By first processing camera images with DriveGuard, we offer a more universal solution than having to re-train each perception model with noisy input. We explore the space of different autoencoder architectures and evaluate them on a diverse dataset created with real and synthetic images demonstrating that by exploiting spatio-temporal information combined with multi-component loss we significantly increase robustness against adverse image effects reaching within 5-6% of that of the original model on clean images.
arXiv comment2021 IEEE Winter Conference on Applications of Computer Vision Workshops (WACVW)
References in corpus (12)
- Rethinking Atrous Convolution for Semantic Image Segmentation
- scikit-image: Image processing in Python
- Adversarial Spatio-Temporal Learning for Video Deblurring
- Image Restoration Using Convolutional Auto-encoders with Symmetric Skip Connections
- Securing Connected & Autonomous Vehicles: Challenges Posed by Adversarial Machine Learning and The Way Forward
- Spatio-temporal video autoencoder with differentiable memory
- NO Need to Worry about Adversarial Examples in Object Detection in Autonomous Vehicles
- SNE-RoadSeg: Incorporating Surface Normal Information into Semantic Segmentation for Accurate Freespace Detection
- Toward Convolutional Blind Denoising of Real Photographs
- Adversarial Scene Editing: Automatic Object Removal from Weak Supervision
- MobilBye: Attacking ADAS with Camera Spoofing
- Talk Proposal: Towards the Realistic Evaluation of Evasion Attacks using CARLA