Adapt visual tasks to new domain without source domain data with DistillAdapt
Adapt visual tasks to new domain without source domain data with DistillAdapt
DistillAdapt: Source-Free Active Visual Domain Adaptation
arXiv paper abstract https://arxiv.org/abs/2205.12840v1
arXiv PDF paper https://arxiv.org/pdf/2205.12840v1.pdf
… present a novel method, DistillAdapt, for the challenging problem of Source-Free Active Domain Adaptation (SF-ADA).
The problem requires adapting a pretrained source domain network to a target domain, within a … budget for acquiring labels in the target domain, while assuming that the source data is not available for adaptation due to privacy concerns or otherwise.
… selective distillation of features from the pre-trained network to the target network using a small subset of annotated target samples mined by H_AL.
… balances transfer-ability from the pre-trained network and uncertainty of the target network.
… is task-agnostic … can be applied across visual tasks such as classification, segmentation and detection. … handle shifts in output label space.
… improvement of 0.5% — 31.3% (across datasets and tasks) over prior adaptation methods that assume access to large amounts of annotated source data for adaptation.
Stay up to date. Subscribe to my posts https://morrislee1234.wixsite.com/website/contact
Web site with my other posts by category https://morrislee1234.wixsite.com/website
LinkedIn https://www.linkedin.com/in/morris-lee-47877b7b
