OurDB: Ouroboric Domain Bridging for Multi-Target Domain Adaptive Semantic Segmentation
This work addresses the challenge of training a single model that performs well across multiple target domains in semantic segmentation, which is crucial for applications like autonomous driving, though it appears incremental as it builds on prior multi-teacher approaches.
The paper tackles multi-target domain adaptation for semantic segmentation by proposing the OurDB framework, which uses a single teacher architecture to dynamically cycle through target domains and incorporates Fisher information to reduce forgetting, achieving superior performance on four urban driving datasets compared to existing state-of-the-art methods.
Multi-target domain adaptation (MTDA) for semantic segmentation poses a significant challenge, as it involves multiple target domains with varying distributions. The goal of MTDA is to minimize the domain discrepancies among a single source and multi-target domains, aiming to train a single model that excels across all target domains. Previous MTDA approaches typically employ multiple teacher architectures, where each teacher specializes in one target domain to simplify the task. However, these architectures hinder the student model from fully assimilating comprehensive knowledge from all target-specific teachers and escalate training costs with increasing target domains. In this paper, we propose an ouroboric domain bridging (OurDB) framework, offering an efficient solution to the MTDA problem using a single teacher architecture. This framework dynamically cycles through multiple target domains, aligning each domain individually to restrain the biased alignment problem, and utilizes Fisher information to minimize the forgetting of knowledge from previous target domains. We also propose a context-guided class-wise mixup (CGMix) that leverages contextual information tailored to diverse target contexts in MTDA. Experimental evaluations conducted on four urban driving datasets (i.e., GTA5, Cityscapes, IDD, and Mapillary) demonstrate the superiority of our method over existing state-of-the-art approaches.