AI Paper Atlas

Research map / Detection and segmentation / Open-set and few-shot detection

Unmasking Anomalies in Road-Scene Segmentation

Shyam Nandan Rai, Fabio Cermelli, Dario Fontanel, Carlo Masone, Barbara Caputo

ICCV 2023 · Detection and segmentation › Open-set and few-shot detection · approach: Industrial anomaly detection · 59 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 global masked attention to Mask2Former (ablation)Mask2Anomaly · FS Lost & Found valAuPRC: 10.60 -> 32.35Improved
use masked attention instead of cross-attention (ablation)Mask2Former · FS Lost & Found valAuPRC: 20.30 -> 10.60Hurt
add mask contrastive learning with outlier masks (ablation)Mask2Anomaly · FS Lost & Found valFPR95: 87.11 -> 31.93 (with RM, with/without GMA)Improved
add refinement mask filtering stuff regions except road (ablation)Mask2Anomaly · FS Lost & Found valFPR95: 24.07 -> 9.46Improved
filter things instead of stuff in refinement mask (ablation)Mask2Anomaly · FS Lost & Found valFPR95: 24.07 -> 39.11Hurt
set contrastive margin to 0.75 rather than 0.5 or 1 (ablation)Mask2Anomaly · FS Lost & Found valAuPRC: 62.07 (m=0.5) -> 69.41Improved
raise batch outlier probability from 0.1 to 0.2 (ablation)Mask2Anomaly · FS Lost & Found valAuPRC: 63.01 -> 69.41Improved
cast anomaly segmentation as mask classification via Mask2Anomaly vs per-pixel Dense HybridMask2Anomaly · Average over five benchmarksFPR95: 16.79 -> 6.68Improved
Mask2Anomaly full model vs vanilla Mask2Former on in-distribution segmentationMask2Anomaly · Cityscapes valmIoU: 78.88 vs 1.46 points above Mask2FormerImproved
use larger backbones for Mask2Former instead of Mask2Anomaly ResNet-50 (ablation)Mask2Former · FS Lost & Found valAuPRC: ResNet-50 10.60 -> Swin-S 30.96 vs Mask2Anomaly 32.35No clear effect
Mask2Anomaly on Road Anomaly versus JSRNetMask2Anomaly · Road AnomalyAuPRC: 94.4 (JSRNet) vs 79.70Hurt

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