AI Paper Atlas

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

Adaptive Texture Filtering for Single-Domain Generalized Segmentation

Xinhui Li, Mingjia Li, Yaxing Wang, Chuan-Xian Ren, Xiaojie Guo

AAAI 2023 · Label-efficient and robust learning › Domain adaptation and generalization · approach: Augmentation and mixture-of-experts generalization · 11 citations (OpenAlex)

Read the paper

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 AFM (ablation)HGGN segmentation model · GTA5 → real-world Avg.mIoU (%): 37.82 -> 44.24Improved
use adaptive filtering strength instead of fixing λa=1 (ablation)AFM · GTA5 → real-world Avg.mIoU (%): 42.11 -> 44.24Improved
use adaptive filtering strength instead of fixing λa=2 (ablation)AFM · GTA5 → real-world Avg.mIoU (%): 40.23 -> 44.24Improved
add HGGN (ablation)AFM segmentation model · GTA5 → real-world Avg.mIoU (%): 38.66 -> 44.24Improved
use 4-level rather than 1-level SGE (ablation)HGGN · GTA5 → real-world Avg.mIoU (%): 40.13 -> 44.24Improved
use 4-level rather than 2-level SGE (ablation)HGGN · GTA5 → real-world Avg.mIoU (%): 41.72 -> 44.24Improved
use 4-level rather than 3-level SGE (ablation)HGGN · GTA5 → real-world Avg.mIoU (%): 42.75 -> 44.24Improved
set α to 2.5 rather than 1.5 (ablation)HGGN contour-supervised objective · GTA5 → real-world Avg.mIoU (%): 41.66 -> 44.24Improved
set α to 2.5 rather than 2.0 (ablation)HGGN contour-supervised objective · GTA5 → real-world Avg.mIoU (%): 43.01 -> 44.24Improved
set α to 2.5 rather than 3.0 (ablation)HGGN contour-supervised objective · GTA5 → real-world Avg.mIoU (%): 43.13 -> 44.24Improved
replace SAN-SAW with the proposed methodResNet-101 segmentation model · GTA5 → CityscapesmIoU (%): 45.33 -> 44.83Hurt
replace SAN-SAW with the proposed methodResNet-50 segmentation model · GTA5 → MapillarymIoU (%): 41.86 -> 42.82Improved

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