AI Paper Atlas

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

Intra- & Extra-Source Exemplar-Based Style Synthesis for Improved Domain Generalization

Yumeng Li, Dan Zhang, Margret Keuper, Anna Khoreva

IJCV 2023 · Label-efficient and robust learning › Domain adaptation and generalization · approach: Domain-generalized segmentation · 11 citations (OpenAlex)

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 random noise masking when training the encoder used for ISSA (ablation)ISSA (HRNet) · CS to Dark ZurichmIoU: 17.36 -> 26.10Improved
add random noise masking when training the encoder used for ISSA (ablation)ISSA (HRNet) · CS to ACDCmIoU: 44.63 -> 47.43Improved
add random noise masking when training the encoder used for ISSA (ablation)ISSA (HRNet) · Cityscapes (source)mIoU: 69.68 -> 69.48No clear effect
raise masking ratio from 25% to 50% (ablation)masked noise encoder (patch 4) · Cityscapes reconstructionFID: 1.41 -> 2.01Hurt
change mask patch size from 2 to 4 (ablation)masked noise encoder (ratio 25%) · Cityscapes reconstructionFID: 1.50 -> 1.41No clear effect
use noise map at 32x64 instead of 4x8 to 8x16 (ablation)masked noise encoder · Cityscapes reconstructionFID: 14.90 -> 2.30Improved
replace Cityscapes-trained encoder with BDD100K-trained encoder and generator (ablation)ISSA (HRNet) · CS to ACDC averagemIoU: 50.1 -> 50.4No clear effect
add extra-source landscape exemplars (ESSA) to ISSAISSA (HRNet) · CS to Dark ZurichmIoU: 27.24 -> 29.06Improved
add ISSA on top of RobustNetRobustNet (DeepLab v3+) · CS to Dark ZurichmIoU: 20.11 -> 23.09Improved
add ISSA on top of SHADESHADE (DeepLab v3+) · CS to Dark ZurichmIoU: 25.37 -> 25.58Mixed
add ISSA on top of MixStyleMixStyle (DeepLab v2) · CS to ACDCmIoU: 36.97 -> 41.81Improved
add StyleMix augmentation instead of baselineHRNet · CS to ACDC-FogmIoU: 58.68 -> 49.11Hurt

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