Alert button

Hierarchical Mask Calibration for Unified Domain Adaptive Panoptic Segmentation

Jun 30, 2022
Jingyi Zhang, Jiaxing Huang, Shijian Lu

Figure 1 for Hierarchical Mask Calibration for Unified Domain Adaptive Panoptic Segmentation
Figure 2 for Hierarchical Mask Calibration for Unified Domain Adaptive Panoptic Segmentation
Figure 3 for Hierarchical Mask Calibration for Unified Domain Adaptive Panoptic Segmentation
Figure 4 for Hierarchical Mask Calibration for Unified Domain Adaptive Panoptic Segmentation

Share this with someone who'll enjoy it:

Domain adaptive panoptic segmentation aims to mitigate data annotation challenge by leveraging off-the-shelf annotated data in one or multiple related source domains. However, existing studies employ two networks for instance segmentation and semantic segmentation separately which lead to a large amount of network parameters with complicated and computationally intensive training and inference processes. We design UniDAPS, a Unified Domain Adaptive Panoptic Segmentation network that is simple but capable of achieving domain adaptive instance segmentation and semantic segmentation simultaneously within a single network. UniDAPS introduces Hierarchical Mask Calibration (HMC) that rectifies the predicted pseudo masks, pseudo superpixels and pseudo pixels and performs network re-training via an online self-training process on the fly. It has three unique features: 1) it enables unified domain adaptive panoptic adaptation; 2) it mitigates false predictions and improves domain adaptive panoptic segmentation effectively; 3) it is end-to-end trainable with much less parameters and simpler training and inference pipeline. Extensive experiments over multiple public benchmarks show that UniDAPS achieves superior domain adaptive panoptic segmentation as compared with the state-of-the-art.

View paper onarxiv icon

Share this with someone who'll enjoy it: