Weakly supervised training of universal visual concepts for multi-domain semantic segmentation
This addresses the challenge of training semantic segmentation models on diverse datasets with inconsistent labels, which is crucial for improving generalization in real-world applications.
The paper tackles the problem of incompatible and overlapping labels across multiple datasets for semantic segmentation by treating labels as unions of universal visual concepts, enabling seamless multi-domain training without relabeling. It achieves competitive or state-of-the-art performance on multi-domain collections and the WildDash 2 benchmark.
Deep supervised models have an unprecedented capacity to absorb large quantities of training data. Hence, training on multiple datasets becomes a method of choice towards strong generalization in usual scenes and graceful performance degradation in edge cases. Unfortunately, different datasets often have incompatible labels. For instance, the Cityscapes road class subsumes all driving surfaces, while Vistas defines separate classes for road markings, manholes etc. Furthermore, many datasets have overlapping labels. For instance, pickups are labeled as trucks in VIPER, cars in Vistas, and vans in ADE20k. We address this challenge by considering labels as unions of universal visual concepts. This allows seamless and principled learning on multi-domain dataset collections without requiring any relabeling effort. 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.