DC Field | Value | Language |
---|---|---|
dc.contributor.author | Shin, Inkyu | ko |
dc.contributor.author | Woo, Sanghyun | ko |
dc.contributor.author | Pan, Fei | ko |
dc.contributor.author | Kweon, In-So | ko |
dc.date.accessioned | 2020-12-16T06:30:26Z | - |
dc.date.available | 2020-12-16T06:30:26Z | - |
dc.date.created | 2020-12-01 | - |
dc.date.created | 2020-12-01 | - |
dc.date.issued | 2020-08 | - |
dc.identifier.citation | European Conference on Computer Vision, ECCV 2020 | - |
dc.identifier.uri | http://hdl.handle.net/10203/278561 | - |
dc.description.abstract | Recently, deep self-training approaches emerged as a powerful solution to the unsupervised domain adaptation. The self-training scheme involves iterative processing of target data; it generates target pseudo labels and retrains the network. However, since only the confident predictions are taken as pseudo labels, existing self-training approaches inevitably produce sparse pseudo labels in practice. We see this is critical because the resulting insufficient training-signals lead to a suboptimal, error-prone model. In order to tackle this problem, we propose a novel Two-phase Pseudo Label Densification framework, referred to as TPLD. In the first phase, we use sliding window voting to propagate the confident predictions, utilizing intrinsic spatial-correlations in the images. In the second phase, we perform a confidence-based easy-hard classification. For the easy samples, we now employ their full pseudo labels. For the hard ones, we instead adopt adversarial learning to enforce hard-to-easy feature alignment. To ease the training process and avoid noisy predictions, we introduce the bootstrapping mechanism to the original self-training loss. We show the proposed TPLD can be easily integrated into existing self-training based approaches and improves the performance significantly. Combined with the recently proposed CRST self-training framework, we achieve new state-of-the-art results on two standard UDA benchmarks. | - |
dc.language | English | - |
dc.publisher | European Conference on Computer Vision | - |
dc.title | Two-Phase Pseudo Label Densification for Self-training based Domain Adaptation | - |
dc.type | Conference | - |
dc.type.rims | CONF | - |
dc.citation.publicationname | European Conference on Computer Vision, ECCV 2020 | - |
dc.identifier.conferencecountry | EI | - |
dc.identifier.conferencelocation | Virtual | - |
dc.contributor.localauthor | Kweon, In-So | - |
dc.contributor.nonIdAuthor | Shin, Inkyu | - |
Items in DSpace are protected by copyright, with all rights reserved, unless otherwise indicated.