Union-set Multi-source Model Adaptation for Semantic Segmentation
This work addresses domain adaptation challenges in semantic segmentation for computer vision applications, offering a more flexible approach to leveraging pre-trained models from multiple sources, though it is incremental in its relaxation of label space constraints.
The paper tackles the problem of multi-source model adaptation for semantic segmentation by relaxing the requirement that each source domain shares a full label space with the target, instead allowing subsets whose union equals the target label space, and proposes a model-invariant feature learning method that improves generalization, achieving superior results in various adaptation settings.
This paper solves a generalized version of the problem of multi-source model adaptation for semantic segmentation. Model adaptation is proposed as a new domain adaptation problem which requires access to a pre-trained model instead of data for the source domain. A general multi-source setting of model adaptation assumes strictly that each source domain shares a common label space with the target domain. As a relaxation, we allow the label space of each source domain to be a subset of that of the target domain and require the union of the source-domain label spaces to be equal to the target-domain label space. For the new setting named union-set multi-source model adaptation, we propose a method with a novel learning strategy named model-invariant feature learning, which takes full advantage of the diverse characteristics of the source-domain models, thereby improving the generalization in the target domain. We conduct extensive experiments in various adaptation settings to show the superiority of our method. The code is available at https://github.com/lzy7976/union-set-model-adaptation.