AI Paper Atlas

Research map / Video / Video-language understanding

DynImg: Key Frames with Visual Prompts are Good Representation for Multi-Modal Video Understanding

Xiaoyi Bao, Chenwei Xie, Hao Tang, Tingyu Weng, Xiaofeng Wang, Yun Zheng, Xingang Wang

ICCV 2025 · Video › Video-language understanding · approach: Long video understanding · 2 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 temporal prompts before the visual encoder (ablation)DynImg · MSVD-QAaccuracy: 74.9 -> 77.3Improved
apply temporal prompts before the encoder instead of after (ablation)DynImg · MSVD-QAaccuracy: 75.2 -> 77.3Improved
add temporal prompts after the encoder (ablation)DynImg · MSVD-QAaccuracy: 74.9 -> 75.2No clear effect
add 4D RoPE (ablation)DynImg · MSVD-QAaccuracy: 77.3 -> 78.6Improved
use 4 N-frames instead of 1 N-frame (ablation)DynImg · MSVD-QAaccuracy: 71.9 -> 78.6Improved
increase N-frames from 4 to 12 (ablation)DynImg · MSVD-QAaccuracy: 78.6 -> 77.5Hurt
use 4 DynImgs instead of 1 (ablation)DynImg · MSVD-QAaccuracy: 77.0 -> 78.6Improved
increase DynImgs from 4 to 16 (ablation)DynImg · MSVD-QAaccuracy: 78.6 -> 77.7Hurt
use 4D coordinates instead of 1D coordinates (ablation)DynImg · MSVD-QAaccuracy: 77.3 -> 78.6Improved
make theta trainable instead of fixed in 4D split RoPE (ablation)DynImg · MSVD-QAaccuracy: 77.0 -> 78.4Improved
merge angles instead of splitting trigonometric terms (ablation)DynImg · MSVD-QAaccuracy: 78.4 -> 78.6Improved
DynImg with 4 DynImgs versus PLLaVA with 16 framesDynImg · average of four open-ended QA setsaccuracy: comparable accuracy with 4 frames vs 16 framesImproved

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