Learning Uncertainty For Safety-Oriented Semantic Segmentation In Autonomous Driving

TL;DR

This paper presents a novel, computationally efficient uncertainty estimation method for semantic segmentation in autonomous driving, enabling safety fallback mechanisms without extra training data.

Contribution

It introduces a new disagreement-based uncertainty measure with a dedicated neural observer trained via self-supervision, improving safety metrics over existing methods.

Findings

Outperforms competing methods in safety evaluations

Less computationally intensive at inference

Effective in glare artifact scenarios

Abstract

In this paper, we show how uncertainty estimation can be leveraged to enable safety critical image segmentation in autonomous driving, by triggering a fallback behavior if a target accuracy cannot be guaranteed. We introduce a new uncertainty measure based on disagreeing predictions as measured by a dissimilarity function. We propose to estimate this dissimilarity by training a deep neural architecture in parallel to the task-specific network. It allows this observer to be dedicated to the uncertainty estimation, and let the task-specific network make predictions. We propose to use self-supervision to train the observer, which implies that our method does not require additional training data. We show experimentally that our proposed approach is much less computationally intensive at inference time than competing methods (e.g. MCDropout), while delivering better results on safety-oriented evaluation metrics on the CamVid dataset, especially in the case of glare artifacts.