Safety Metrics for Semantic Segmentation in Autonomous Driving
This work addresses safety evaluation for autonomous driving systems, but it is incremental as it extends existing safety metrics from classification and detection to segmentation.
The paper tackles the problem of evaluating safety in semantic segmentation for autonomous driving by proposing new metrics that consider spatial clustering and location of classification errors, demonstrating their validity and practicality on an autonomous driving dataset.
Within the context of autonomous driving, safety-related metrics for deep neural networks have been widely studied for image classification and object detection. In this paper, we further consider safety-aware correctness and robustness metrics specialized for semantic segmentation. The novelty of our proposal is to move beyond pixel-level metrics: Given two images with each having N pixels being class-flipped, the designed metrics should, depending on the clustering of pixels being class-flipped or the location of occurrence, reflect a different level of safety criticality. The result evaluated on an autonomous driving dataset demonstrates the validity and practicality of our proposed methodology.