AI Paper Atlas

Research map / Video / Video-language understanding

CVA: Context-aware Video-text Alignment for Video Temporal Grounding

Sungho Moon, Seunghun Lee, Jiwan Seo, Sunghoon Im

CVPR 2026 · Video › Video-language understanding · approach: Video-language pretraining · 1 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 query-aware context diversification (QCD) (ablation)baseline · QVHighlights valR1@0.7: 46.77 -> 51.98Improved
use QCD instead of query-agnostic mix (ablation)baseline · QVHighlights valR1@0.7: 48.39 -> 51.98Improved
add CTE on top of QCD (ablation)QCD model · QVHighlights valR1@0.7: 51.98 -> 52.63Improved
add CBD loss on top of QCD (ablation)QCD model · QVHighlights valR1@0.7: 51.98 -> 53.02Improved
add all three components (ablation)baseline · QVHighlights valR1@0.7: 46.77 -> 54.84Improved
add semantically hard negatives to CBD (Nhard 0 -> 2) (ablation)CBD · QVHighlights valR1@0.7: 51.94 -> 54.89Improved
increase Nhard from 5 to 10 (ablation)CBD · QVHighlights valR1@0.7: 54.84 -> 52.71Hurt
temporal-only negatives in CBD (Nhard=0) (ablation)CBD · QVHighlights valR1@0.7: 51.98 -> 51.94No clear effect
raise upper percentile to 90 (ablation)QCD · QVHighlights valR1@0.7: 50.92 -> 48.88Hurt
raise lower percentile to 40 (ablation)QCD · QVHighlights valR1@0.7: 50.92 -> 49.71Hurt
add learnable queries only to CTE (ablation)QCD model · QVHighlights valR1@0.7: 51.98 -> 52.33No clear effect
add windowed self-attention only to CTE (ablation)QCD model · QVHighlights valR1@0.7: 51.98 -> 52.13No 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