AI Paper Atlas

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

Towards Reliable Evaluation and Fast Training of Robust Semantic Segmentation Models

Francesco Croce, Naman D. Singh, Matthias Hein

ECCV 2024 · Adversarial robustness and security › Attacks on text and graphs · approach: Biologically inspired robust vision · 14 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
initialize backbone with an l_inf-robust ImageNet classifier (PIR-AT) instead of clean init (ablation)UPerNet+ConvNeXt-T adversarial training, 2 steps, 50 epochs · Pascal-VOC eps 8/255robust pixel accuracy (%): 0.0 -> 50.2Improved
use robust ImageNet init with 5 attack steps (ablation)UPerNet+ConvNeXt-T AT, 50 epochs · Pascal-VOC eps 8/255robust pixel accuracy (%): 65.3 -> 71.7Improved
use robust ImageNet init (PIR-AT) with 5 steps (ablation)UPerNet+ConvNeXt-T AT, 128 epochs · ADE20K eps 4/255robust mIoU (%): 12.8 -> 17.2Improved
use robust ImageNet init (PIR-AT) with 5 steps (ablation)Segmenter+ViT-S AT, 128 epochs · ADE20K eps 8/255robust pixel accuracy (%): 25.1 -> 33.3Improved
reduce PIR-AT training epochs from 128 to 32 (ablation)UPerNet+ConvNeXt-T on ADE20K · ADE20K eps 8/255robust pixel accuracy (%): 26.4 -> 30.6Improved
use progressive radius reduction (red-eps) instead of constant radius in APGD (ablation)SEA attacks · Pascal-VOC PIR-AT UPerNet+ConvNeXt-Trobust accuracyImproved
increase SEA iterations from 300 to 500 (ablation)SEA · Pascal-VOC eps 8/255 and 12/255robust accuracy and mIoU: around 0.1% dropNo clear effect
ensemble JS, MCE and MCE-Bal losses (SEA) instead of any pair (ablation)SEA · Pascal-VOC eps 12/255robust pixel accuracy (%): 29.4 -> 28.1Improved
use SEA instead of SegPGD for evaluationRobust segmentation models · Pascal-VOC and ADE20K, six modelsrobust accuracy: up to 17.2% lower accuracy and 10.3% lower mIoUImproved
use JS, MCE or MCE-Bal loss instead of CE-based attacksAPGD attack, 300 iterations · Pascal-VOC and ADE20Krobust accuracyImproved
use the segmentation-specific losses (JS etc.) for adversarial training instead of cross-entropy (ablation)PIR-AT / AT · Pascal-VOC, ADE20KrobustnessNo clear effect
train with only 2 attack steps and clean init for 50 epochs (ablation)UPerNet+ConvNeXt-T AT · Pascal-VOC eps 4/255robust pixel accuracy (%): 2.6 (vs 86.7 for 2-step PIR-AT)Hurt

Earlier methods this paper compares with

Named baselines from earlier accepted papers in this paper's results tables: better than 1 of 1 on most shared results.

Matched by method name in results tables read by AI. Authors' own tables tend to favour the newer method, and a comparison can be missed when a method is renamed.

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