AI Paper Atlas

Research map / Detection and segmentation / CNN features

Energy-Based Models for Deep Probabilistic Regression

Fredrik K. Gustafsson, Martin Danelljan, Goutam Bhat, Thomas B. Schön

ECCV 2020 · Detection and segmentation › CNN features · approach: Efficient DETR query design · 63 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
replace direct regression with energy-based modelFaster-RCNN box regression · COCO 2017 test-devAP (%): 37.2 -> 39.4Improved
replace confidence (IoU) regression with energy-based NLL trainingIoU-Net · COCO 2017 test-devAP (%): 38.3 -> 39.4Improved
use same proposal and inference settings for IoU-NetIoU-Net · COCO 2017 test-devAP (%): 38.3 -> 38.2No clear effect
use simple Gaussian output model instead of direct regressionFaster-RCNN · COCO 2017 test-devAP (%): 37.2 -> 36.7Hurt
add more mixture components (2, 4, 8)Gaussian mixture · COCO 2017 test-devAP (%): 37.1 -> 36.8No clear effect
replace ATOM IoU-Net module with energy-based regressionATOM tracker · TrackingNetSuccess (%): 70.3 -> 74.5Improved
replace ATOM IoU-Net module with energy-based regressionATOM tracker · UAV123AUC (%): 65.0 -> 67.2Improved
add gradient-based refinementDirect L2 regression · UTKFace age estimationMAE: 4.81 -> 4.65Improved
add gradient-based refinementLaplace · UTKFace age estimationMAE: 4.85 -> 4.81No clear effect
add gradient-based refinementDirect · BIWI head poseaverage MAE: 3.09 -> 3.07No clear effect
use two-component Gaussian proposal instead of one (ablation)proposal distribution · COCO 2017 valAP (%): 38.47 -> 39.05Improved
use three components instead of two (ablation)proposal distribution · COCO 2017 valAP (%): 39.05 -> 39.11No clear effect

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