AI Paper Atlas

Research map / Video / Video segmentation and tracking

Learning Flexible Generalization in Video Quality Assessment by Bringing Device and Viewing Condition Distributions

Nikolay Safonov, Dmitriy S. Vatolin

ICML 2026 · Video › Video segmentation and tracking · approach: Video restoration with temporal priors · 0 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 viewing-condition adaptation module on top of LPIPSLPIPS · held-out phone model 1Kendall rank correlation: 0.308 -> 0.652Improved
add viewing-condition adaptation module on top of VMAFVMAF · held-out phone model 1Kendall rank correlation: 0.509 -> 0.600Improved
add adaptation module on top of AVQTAVQT · held-out phone model 1Kendall rank correlation: 0.439 -> 0.694Improved
add adaptation module on top of PSNRPSNR · held-out phone model 3Kendall rank correlation: 0.501 -> 0.430Hurt
add adaptation module on top of NLPDNLPD · held-out phone models 2 and 3Kendall rank correlation: 0.468 -> 0.415Hurt
add adaptation module on top of MUSIQMUSIQ · held-out phone models 1-3Kendall rank correlation: 0.320 -> 0.320No clear effect
add adaptation module to all metrics, evaluated on held-out devices32 IQA/VQA metrics · three most frequent phone modelsKendall rank correlationMixed
use condition pool: sample viewing conditions independently of annotations to augment training (ablation)adaptation module · held-out devicesKendall rank correlationImproved
use Blade-Chest aggregation conditioned on viewing conditions instead of Bradley-Terryscore aggregation · dataset scoresSROCC: 0.881 (vs BT-scores)No clear effect
use tanh activations in networks instead of linear (ablation)Blade-Chest networks · datasetscore fitImproved
evaluate on HDR-capable vs SDR displays for HDR videosubjective scores · HDR videosPLCC vs reference: 0.881 -> 0.933Improved
larger screen size group of devicescrowd subjective scores · agreement with desktop scoresPLCC: 0.909 -> 0.945Improved

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