Domain Adaptive Knowledge Distillation for Driving Scene Semantic Segmentation
This addresses domain adaptation for semantic segmentation in autonomous driving, but it is incremental as it builds on existing knowledge distillation techniques.
The paper tackles memory constraints and domain gaps in autonomous driving systems by proposing Domain Adaptive Knowledge Distillation, a method that uses multi-level distillation and a novel cross entropy loss with pseudo labels, achieving success in real-to-real and synthetic-to-real scenarios.
Practical autonomous driving systems face two crucial challenges: memory constraints and domain gap issues. In this paper, we present a novel approach to learn domain adaptive knowledge in models with limited memory, thus bestowing the model with the ability to deal with these issues in a comprehensive manner. We term this as "Domain Adaptive Knowledge Distillation" and address the same in the context of unsupervised domain-adaptive semantic segmentation by proposing a multi-level distillation strategy to effectively distil knowledge at different levels. Further, we introduce a novel cross entropy loss that leverages pseudo labels from the teacher. These pseudo teacher labels play a multifaceted role towards: (i) knowledge distillation from the teacher network to the student network & (ii) serving as a proxy for the ground truth for target domain images, where the problem is completely unsupervised. We introduce four paradigms for distilling domain adaptive knowledge and carry out extensive experiments and ablation studies on real-to-real as well as synthetic-to-real scenarios. Our experiments demonstrate the profound success of our proposed method.