A benchmark of 32 image perturbations on two ADAS shows most corruptions cause failures, and retraining on perturbed data improves robustness to simulated weather, but the retraining effect is confounded by new road data.
Generalisation in humans and deep neural networks,
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
fields
cs.SE 1years
2025 1verdicts
REJECT 1representative citing papers
citing papers explorer
-
Benchmarking Image Perturbations for Testing Automated Driving Assistance Systems
A benchmark of 32 image perturbations on two ADAS shows most corruptions cause failures, and retraining on perturbed data improves robustness to simulated weather, but the retraining effect is confounded by new road data.