Evaluation of Out-of-Distribution Detection Performance on Autonomous Driving Datasets. (arXiv:2401.17013v1 [cs.LG])

Safety measures need to be systemically investigated to what extent they
evaluate the intended performance of Deep Neural Networks (DNNs) for critical
applications. Due to a lack of verification methods for high-dimensional DNNs,
a trade-off is needed between accepted performance and handling of
out-of-distribution (OOD) samples.

This work evaluates rejecting outputs from semantic segmentation DNNs by
applying a Mahalanobis distance (MD) based on the most probable
class-conditional Gaussian distribution for the predicted class as an OOD
score. The evaluation follows …

applications arxiv autonomous autonomous driving cs.lg datasets detection distribution driving evaluation networks neural networks performance safety safety measures samples trade trade-off verification work