AI Paper Atlas

Research map / Detection and segmentation / CNN features

Towards Robust Semantic Segmentation against Patch-Based Attack via Attention Refinement

Zheng Yuan, Jie Zhang, Yude Wang, Shiguang Shan, Xilin Chen

IJCV 2024 · Detection and segmentation › CNN features · approach: Context modeling for semantic segmentation · 6 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 Max Attention Suppression (MAS) (ablation)SegFormer/MiT · ADE20K ablation (DAG attack, SegFormer/MiT)Permute target mIoU: 49.32 -> 34.11Improved
add Random Attention Dropout (RAD) (ablation)SegFormer/MiT · ADE20K ablation (DAG attack, SegFormer/MiT)Permute target mIoU: 49.32 -> 46.77Improved
add RAD on top of MAS (ablation)SegFormer/MiT · ADE20K ablation (DAG attack, SegFormer/MiT)Strip target mIoU: 23.36 -> 22.63Improved
add MAS and RAD to baseline (ablation)SegFormer/MiT · ADE20K ablation (DAG attack, SegFormer/MiT)clean mIoU: 48.25 -> 48.30No clear effect
lower MAS threshold T from 0.6 to 0.3 (ablation)SegFormer/MiT · ADE20K (DAG attack)Permute target mIoU: 37.69 -> 34.11Improved
lower MAS threshold T from 0.3 to 0.2 (ablation)SegFormer/MiT · ADE20K (DAG attack)Permute target mIoU: 34.11 -> 35.12Hurt
set RAD dropout rate to 0.5 versus 0.4 (ablation)SegFormer/MiT · ADE20K (DAG attack)Permute target mIoU: 48.51 -> 46.77Improved
add RAM (vs baseline) to Nonlocal/R50Nonlocal/R50 · ADE20K (DAG attack)Permute target mIoU: 40.42 -> 14.59Improved
add RAM (vs baseline) to UPerNet/DeiTUPerNet/DeiT · ADE20K (DAG attack)Strip target mIoU: 85.11 -> 48.27Improved
use RAM instead of Learnable scaling on Segmenter/ViTSegmenter/ViT · ADE20K (DAG attack)Permute target mIoU: 75.50 -> 53.76Improved
use RAM instead of adversarial training (DAG attack)SegFormer/MiT · ADE20K (DAG targeted attack, 150x150 patch)Permute target mIoU: 39.72 -> 33.72Improved
combine RAM with DDC-ATNonlocal/R50 · Cityscapes (DAG untargeted attack, 100x100 patch)mIoU under attack: 34.78 -> 37.81Improved

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