arXiv Analytics

Sign in

arXiv:1708.03309 [cs.CV]AbstractReferencesReviewsResources

Systematic Testing of Convolutional Neural Networks for Autonomous Driving

Tommaso Dreossi, Shromona Ghosh, Alberto Sangiovanni-Vincentelli, Sanjit A. Seshia

Published 2017-08-10Version 1

We present a framework to systematically analyze convolutional neural networks (CNNs) used in classification of cars in autonomous vehicles. Our analysis procedure comprises an image generator that produces synthetic pictures by sampling in a lower dimension image modification subspace and a suite of visualization tools. The image generator produces images which can be used to test the CNN and hence expose its vulnerabilities. The presented framework can be used to extract insights of the CNN classifier, compare across classification models, or generate training and validation datasets.

Related articles: Most relevant | Search more
arXiv:1505.00256 [cs.CV] (Published 2015-05-01)
DeepDriving: Learning Affordance for Direct Perception in Autonomous Driving
arXiv:1607.00971 [cs.CV] (Published 2016-07-04)
Can we unify monocular detectors for autonomous driving by using the pixel-wise semantic segmentation of CNNs?
arXiv:1612.07695 [cs.CV] (Published 2016-12-22)
MultiNet: Real-time Joint Semantic Reasoning for Autonomous Driving