Kundu, JN and Kulkarni, A and Singh, A and Jampani, V and Babu, RV (2021) Generalize then Adapt: Source-Free Domain Adaptive Semantic Segmentation. In: 18th IEEE/CVF International Conference on Computer Vision, ICCV 2021, 1 - 17 October 2021, Virtual, Online, pp. 7026-7036.
|
PDF
IEEE_ICCV_2021.pdf - Published Version Download (5MB) | Preview |
Abstract
Unsupervised domain adaptation (DA) has gained substantial interest in semantic segmentation. However, almost all prior arts assume concurrent access to both labeled source and unlabeled target, making them unsuitable for scenarios demanding source-free adaptation. In this work, we enable source-free DA by partitioning the task into two: a) source-only domain generalization and b) source-free target adaptation. Towards the former, we provide theoretical insights to develop a multi-head framework trained with a virtually extended multi-source dataset, aiming to balance generalization and specificity. Towards the latter, we utilize the multi-head framework to extract reliable target pseudo-labels for self-training. Additionally, we introduce a novel conditional prior-enforcing auto-encoder that discourages spatial irregularities, thereby enhancing the pseudo-label quality. Experiments on the standard GTA5�Cityscapes and SYNTHIA�Cityscapes benchmarks show our superiority even against the non-source-free prior-arts. Further, we show our compatibility with online adaptation enabling deployment in a sequentially changing environment. © 2021 IEEE
| Item Type: | Conference Proceedings |
|---|---|
| Publication: | Proceedings of the IEEE International Conference on Computer Vision |
| Publisher: | Institute of Electrical and Electronics Engineers Inc. |
| Additional Information: | The copyright for this article belongs to Authors. |
| Department/Centre: | Division of Electrical Sciences > Computer Science & Automation |
| Date Deposited: | 19 May 2022 10:36 |
| Last Modified: | 19 May 2022 10:36 |
| URI: | https://eprints.iisc.ac.in/id/eprint/71896 |
Actions (login required)
 |
View Item |
