AI Paper Atlas

Research map / Video / Video-language understanding

Breaking the Encoder Barrier for Seamless Video-Language Understanding

Handong Li, Yiyuan Zhang, Longteng Guo, Xiangyu Yue, Jing Liu

ICCV 2025 · Video › Video-language understanding · approach: Video LLMs with fine-grained understanding · 6 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 cross-attention video patch embedding layer (ablation)ELVA 7B · VideoMMEaccuracy: 39.6 -> 42.2Improved
add cross-attention video patch embedding layer (ablation)ELVA 7B · MSVD-QAaccuracy: 48.7 -> 49.9Improved
use video-pretrained SigLIP as guidance target (ablation)ELVA 7B · VideoMMEaccuracy: 41.0 -> 42.2Improved
use video-pretrained EVACLIP-g as guidance target (ablation)ELVA 7B · VideoMMEaccuracy: 36.8 -> 38.4Improved
replace original captions with recaptioned image+video data (ablation)ELVA 7B · VideoMMEaccuracy: 34.2 -> 38.5Improved
recaption data (ablation)ELVA 7B · GQAaccuracy: 42.1 -> 46.1Improved
scale pretraining data from 1M to 7M (ablation)ELVA 7B · MSVD-QAaccuracy: 47.1 -> 49.4Improved
scale pretraining data from 1M to 7M (ablation)ELVA 7B · VideoMMEaccuracy: 38.1 -> 38.5Mixed
add MSE tube-wise alignment loss to generative loss (ablation)ELVA 7B · GQAaccuracy: 42.2 -> 43.6Improved
add contrastive frame-wise loss to generative loss (ablation)ELVA 7B · VideoMMEaccuracy: 37.9 -> 38.5Improved
combine MSE and contrastive losses with generative loss (ablation)ELVA 7B · SEED-Iaccuracy: 40.0 -> 44.8Improved
add 16 low-res frames to 16 high-res frames at inference (ablation)ELVA 1.5B · VideoMMEaccuracy: 37.1 -> 43.0Improved

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