False Negative Reduction in Semantic Segmentation under Domain Shift using Depth Estimation
This work addresses safety-critical applications like automated driving by reducing false negatives in semantic segmentation when models face domain shifts, though it is incremental as it post-processes existing methods.
The paper tackles the problem of false negatives in semantic segmentation under domain shift by using depth estimation to generate foreground-background masks and aggregating them with segmentation outputs, resulting in reduced non-detected objects for key classes and improved generalization across domains.
State-of-the-art deep neural networks demonstrate outstanding performance in semantic segmentation. However, their performance is tied to the domain represented by the training data. Open world scenarios cause inaccurate predictions which is hazardous in safety relevant applications like automated driving. In this work, we enhance semantic segmentation predictions using monocular depth estimation to improve segmentation by reducing the occurrence of non-detected objects in presence of domain shift. To this end, we infer a depth heatmap via a modified segmentation network which generates foreground-background masks, operating in parallel to a given semantic segmentation network. Both segmentation masks are aggregated with a focus on foreground classes (here road users) to reduce false negatives. To also reduce the occurrence of false positives, we apply a pruning based on uncertainty estimates. Our approach is modular in a sense that it post-processes the output of any semantic segmentation network. In our experiments, we observe less non-detected objects of most important classes and an enhanced generalization to other domains compared to the basic semantic segmentation prediction.