AI Paper Atlas

Research map / Adversarial robustness and security / Attacks on text and graphs

Dynamic Divide-and-Conquer Adversarial Training for Robust Semantic Segmentation

Xiaogang Xu, Hengshuang Zhao, Jiaya Jia

ICCV 2021 · Adversarial robustness and security › Attacks on text and graphs · approach: Biologically inspired robust vision · 38 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 standard adversarial training (SAT)PSPNet · VOC, white-box BIM L_inf attackmIoU BIM n=6: 5.4 -> 27.9Improved
add standard adversarial training (SAT)DeepLabv3 · VOC, white-box BIM L_inf attackmIoU BIM n=6: 5.5 -> 28.8Improved
add standard adversarial training (SAT)PSPNet · VOC, clean samplesmIoU: 76.9 -> 74.3Hurt
replace SAT with DDC-ATPSPNet · VOC, clean samplesmIoU: 74.3 -> 76.0Improved
replace SAT with DDC-ATDeepLabv3 · VOC, clean samplesmIoU: 72.7 -> 75.2Improved
replace SAT with DDC-ATPSPNet · VOC, white-box BIM L_inf attackmIoU BIM n=6: 27.9 -> 33.8Improved
replace SAT with DDC-ATDeepLabv3 · VOC, white-box BIM L_inf attackmIoU BIM n=6: 28.8 -> 32.3Improved
replace SAT with DDC-ATPSPNet · Cityscapes, white-box BIM L_inf attackmIoU BIM n=6: 25.8 -> 28.7Improved
replace SAT with DDC-ATDeepLabv3 · Cityscapes, white-box BIM L_inf attackmIoU BIM n=6: 26.2 -> 29.0Improved
replace SAT with DDC-ATPSPNet · Cityscapes, clean samplesmIoU: 69.0 -> 71.7Improved
use DDC-AT division with clean boundary pixels in auxiliary branch (DDC-AT-M) instead of the full dynamic division (ablation)DDC-AT · VOC ablation, PSPNet, BIM n=6mIoU: 33.8 -> 30.4Hurt
use DDC-AT-N division instead of the full DDC-AT division (ablation)DDC-AT · VOC ablation, PSPNet, BIM n=6mIoU: 33.8 -> 31.8Hurt

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