AI Paper Atlas

Research map / Detection and segmentation / CNN features

Uncertainty Quantification in Detection Transformers: Object-Level Calibration and Image-Level Reliability

Young‐Jin Park, Carson Sobolewski, Navid Azizan

TPAMI 2026 · Detection and segmentation › CNN features · approach: Box regression and loss design · 3 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 negative-prediction confidence with weight lambda (contrast) to positive confidence (ablation)Conf+ (OCE) · COCO, Cal-DETRPearson correlation with image reliability: 0.547 -> 0.628Improved
add contrast with negatives (ablation)Conf+ (OCE) · COCO, UP-DETRPearson correlation with image reliability: 0.477 -> 0.656Improved
add contrast with negatives (ablation)Conf+ (OCE) · Foggy Cityscapes, D-DETRPearson correlation with image reliability: 0.276 -> 0.350Improved
select positives with OCE-minimizing post-processing instead of threshold 0.3Conf+ (0.3) · Foggy Cityscapes, UP-DETRPearson correlation with image reliability: 0.195 -> 0.191No clear effect
use optimal matching instead of OCE-based separationContrastiveConf · COCO, UP-DETRPearson correlation with image reliability: 0.656 -> 0.700Improved
use top-100 predictions as positivesConf+ · COCO, UP-DETRPearson correlation with image reliability: 0.476 -> -0.619Hurt
use lambda between 1.0 and 10.0 instead of 0.0 in ContrastiveConf (ablation)ContrastiveConf (OCE) · COCOPearson correlation with image reliabilityImproved
use excessively large lambda (ablation)ContrastiveConf (OCE) · Foggy CityscapesPearson correlation with image reliabilityHurt
replace averaging with Max-IoU or IoU-weighted aggregation in OCE (ablation)OCE · 6 models x 3 distributionscorrelation between OCE variantsNo clear effect
vary IoU threshold tau in OCE (ablation)OCE · 6 models x 3 distributionscorrelation between OCE variantsNo clear effect

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