Mitigating Distributional Shift in Semantic Segmentation via Uncertainty Estimation from Unlabelled Data
This addresses safety concerns in autonomous vehicles by improving error detection in unseen domains, but it is incremental as it builds on existing uncertainty estimation techniques.
The paper tackles the problem of distributional shift in semantic segmentation for autonomous vehicles by proposing a method that uses unlabelled data to estimate uncertainty and detect errors without additional annotation, achieving improvements of up to 10.7% in ROC-AUC and 19.2% in PR-AUC on a challenging benchmark.
Knowing when a trained segmentation model is encountering data that is different to its training data is important. Understanding and mitigating the effects of this play an important part in their application from a performance and assurance perspective - this being a safety concern in applications such as autonomous vehicles (AVs). This work presents a segmentation network that can detect errors caused by challenging test domains without any additional annotation in a single forward pass. As annotation costs limit the diversity of labelled datasets, we use easy-to-obtain, uncurated and unlabelled data to learn to perform uncertainty estimation by selectively enforcing consistency over data augmentation. To this end, a novel segmentation benchmark based on the SAX Dataset is used, which includes labelled test data spanning three autonomous-driving domains, ranging in appearance from dense urban to off-road. The proposed method, named Gamma-SSL, consistently outperforms uncertainty estimation and Out-of-Distribution (OoD) techniques on this difficult benchmark - by up to 10.7% in area under the receiver operating characteristic (ROC) curve and 19.2% in area under the precision-recall (PR) curve in the most challenging of the three scenarios.