Multi-domain semantic segmentation with overlapping labels
This addresses the challenge of graceful degradation in unusual scenes for semantic segmentation applications, but it is incremental as it builds on existing multi-dataset training approaches.
The paper tackles the problem of training semantic segmentation models on multiple datasets with incompatible and overlapping labels by proposing a method based on partial labels and probabilistic loss, achieving competitive or state-of-the-art performance on multi-domain dataset collections and the WildDash 2 benchmark.
Deep supervised models have an unprecedented capacity to absorb large quantities of training data. Hence, training on many datasets becomes a method of choice towards graceful degradation in unusual scenes. Unfortunately, different datasets often use incompatible labels. For instance, the Cityscapes road class subsumes all driving surfaces, while Vistas defines separate classes for road markings, manholes etc. We address this challenge by proposing a principled method for seamless learning on datasets with overlapping classes based on partial labels and probabilistic loss. Our method achieves competitive within-dataset and cross-dataset generalization, as well as ability to learn visual concepts which are not separately labeled in any of the training datasets. Experiments reveal competitive or state-of-the-art performance on two multi-domain dataset collections and on the WildDash 2 benchmark.