AI Paper Atlas

Research map / Label-efficient and robust learning / Domain adaptation and generalization

Self-Mutating Network for Domain Adaptive Segmentation in Aerial Images

Kyungsu Lee, Haeyun Lee, Jae Youn Hwang

ICCV 2021 · Label-efficient and robust learning › Domain adaptation and generalization · approach: Domain-generalized segmentation

Read the paperCheck your idea against 137,217 papers

AI-extracted. The measured effects below were read by AI from the paper's full text (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.

Measured effects reported in the paper

What was changedWhereMeasuredEffect
add parameter mutation (GAN-based test-time tuning) to baseline (ablation)SegNet baseline · Inria-trained to other domainsbuilding IoU (%): 49.2 -> 52.8Improved
add parameter fluctuation to baseline (ablation)SegNet baseline · Inria-trained to other domainsbuilding IoU (%): 49.2 -> 52.9Improved
add parameter fluctuation on top of parameter mutation (ablation)SMN · Inria-trained to other domainsbuilding IoU (%): 52.8 -> 62.9Improved
add parameter mutation on top of parameter fluctuation (ablation)SMN · Massachusetts-trained to other domainsbuilding IoU (%): 52.4 -> 61.6Improved
generate similar-but-not-identical images (structural similarity weight) instead of identical (SAME) (ablation)SMN · Inria-trained to other domainsbuilding IoU (%): 52.2 -> 62.9Improved
generate similar-but-not-identical images instead of SAME, WHU-trained (ablation)SMN · WHU-trained to other domainsbuilding IoU (%): 53.2 -> 61.9Improved
SMN versus best domain-adaptation baseline, trained on InriaSMN · Inria average mean IoUmean IoU: 0.6193 -> 0.6290Improved
SMN versus best baseline, trained on OUDSMN · OUD average mean IoUmean IoU: 0.5973 -> 0.6086Improved
train on two domains rather than oneSMN · cross-domain building segmentationIoU: 0.643 -> 0.717Improved

Closest papers by meaning

Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.

Open scime draft

Get the next scime Dispatch and a monthly field update

Free. We send a confirmation link first; every email has a one-click unsubscribe. Privacy