AI Paper Atlas

Research map / Label-efficient and robust learning / OOD and anomaly detection

Out-of-Distribution Identification: Let Detector Tell Which I Am Not Sure

Ruoqi Li, Chongyang Zhang, Hao Zhou, Chao Shi, Yan Luo

ECCV 2022 · Label-efficient and robust learning › OOD and anomaly detection · approach: Provable and feature-based OOD detection · 7 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 Attractive-Repulsive loss for feature structuration (ablation)FOOD-ID-Proto (FCOS) · PASCAL VOC-IOAUC: 0.705 -> 0.899Improved
add Attractive-Repulsive loss for feature structuration (ablation)FOOD-ID-Proto (Faster RCNN) · PASCAL VOC-IOAUC: 0.687 -> 0.861Improved
add Attractive-Repulsive loss for feature structuration (ablation)FOOD-ID-CCNF (FCOS) · PASCAL VOC-IOAUC: 0.888 -> 0.913Improved
add Attractive-Repulsive loss for feature structuration (ablation)FOOD-ID-CCNF (RetinaNet) · PASCAL VOC-IOAUC: 0.921 -> 0.928Improved
use class-conditional normalizing flow instead of unconditional flows per class (ablation)FOOD-ID (FCOS) · PASCAL VOC-IOAUC: 0.881 -> 0.913Improved
use class-conditional normalizing flow instead of unconditional flows per class (ablation)FOOD-ID (Faster RCNN) · PASCAL VOC-IOTPR@5%FPR: 39.2 -> 57.7Improved
use CCNF likelihood instead of prototype distance for uncertaintyFOOD-ID (FCOS) · PASCAL VOC-IOAUC: 0.899 -> 0.913Improved
use FOOD-ID-CCNF instead of detector confidence score for OOD identificationFCOS · PASCAL VOC-IOTPR@5%FPR: 43.5 -> 70.1Improved
use FOOD-ID-CCNF instead of detector confidence score for OOD identificationFaster RCNN · Crack DefectAUC: 0.872 -> 0.908Improved
set beta to 0 in CCNF loss (no class-conditioning) (ablation)FOOD-ID-CCNF (FCOS) · PASCAL VOC-IOAUC: 0.913 -> 0.500Hurt
vary beta from 0.5 to 10 in CCNF loss (ablation)FOOD-ID-CCNF (FCOS) · PASCAL VOC-IOAUC: 0.911 -> 0.912No clear effect
exclude 10% of detections identified as OOD with FOOD-ID-CCNFFCOS · PASCAL VOC-IOprecision of IND set: 0.797 -> 0.836Improved

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