AI Paper Atlas

Research map / Detection and segmentation / CNN features

"On Calibration of Object Detectors: Pitfalls, Evaluation and Baselines"

Selim Kuzucu, Kemal Oksuz, Jonathan Sadeghi, Puneet Dokania

ECCV 2024 · Detection and segmentation › CNN features · approach: Box regression and loss design · 18 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
fit post-hoc calibrator on in-distribution val set instead of domain-shifted Objects365 (ablation)Isotonic Regression on D-DETR · COCO minitestD-ECE: 14.2 -> 1.3Improved
train D-DETR on Cityscapes with 4x longer schedule and larger images (ablation)D-DETR vs Cal-DETR baselines · CityscapesD-ECE: 13.8 -> 2.9Improved
replace train-time calibration (Cal-DETR) with post-hoc Isotonic RegressionD-DETR · COCO minitestLaECE0: 11.6 -> 7.7Improved
replace train-time calibration (Cal-DETR) with post-hoc Isotonic RegressionD-DETR · Cityscapes minitestLaECE0: 21.3 -> 9.0Improved
post-hoc calibration under large natural domain shiftIsotonic Regression on D-DETR · Obj45KLaECE0: 16.3 -> 17.2Hurt
use domain-shifted val set for TS thresholds and calibrator (ablation)Temperature Scaling · COCO minitestLRP: 57.3 -> 61.5Hurt
add detection thresholding and class-wise calibrators to Temperature Scaling (ablation)Temperature Scaling · Cityscapes minitestLaECE0: 20.9 -> 12.7Improved
add bias term (Platt Scaling instead of TS) (ablation)Temperature Scaling · Cityscapes minitestLaECE0: 12.7 -> 9.6Improved
optimize calibrator for the evaluation measure (D-ECE vs LaECE objective) (ablation)Platt Scaling · COCO minitestLaECE: PS for D-ECE: LaECE 16.3 vs PS for LaECE: 11.5Mixed
calibrate 14 diverse detectors with Isotonic Regressionvarious detectors · COCO minitestLaECE0: 10.6-34.4 -> 6.7-8.9Improved
use Isotonic Regression instead of Platt Scaling (ablation)post-hoc calibrators · COCO minitestLaECE0: 9.6 -> 7.7Improved
apply Cal-DETR style train-time regularization (MbLS) for calibrationD-DETR · COCO minitestD-ECE: 12.8 -> 15.6Hurt

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