AI Paper Atlas

Research map / Video / Video-language understanding

B-VLLM: A Vision Large Language Model with Balanced Spatio-Temporal Tokens

Zhuqiang Lu, Zhenfei Yin, Mengwei He, Zhihui Wang, Zicheng Liu, Zhiyong Wang, Kun Hu

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 adaptive frame selection (ablation)LLaMA-VID · VideoMMEaccuracy: 33.6 -> 36.5Improved
add adaptive frame selection (ablation)LLaMA-VID · MMBenchaccuracy: 49.8 -> 58.0Improved
add temporal frame token merging on top of frame selection (ablation)LLaMA-VID · MVBenchaccuracy: 39.4 -> 42.1Improved
add temporal frame token merging on top of frame selection (ablation)LLaMA-VID · POPEaccuracy: 81.1 -> 67.3Hurt
add spatial visual token sampling on top of F.S. and T.M. (ablation)LLaMA-VID · MVBenchaccuracy: 42.1 -> 43.5Improved
add spatial visual token sampling on top of F.S. and T.M. (ablation)LLaMA-VID · POPEaccuracy: 67.3 -> 83.8Improved
integrate the method into LLaMA-VIDLLaMA-VID · VideoMME longaccuracy: 27.1 -> 35.2Improved
integrate the method into Video-LLaMA2Video-LLaMA2 · VideoMME longaccuracy: 37.7 -> 41.5Improved
integrate the method into Video-LLaMA2Video-LLaMA2 · VideoMME shortaccuracy: 48.9 -> 47.2Hurt
integrate the method into LLaMA-VIDLLaMA-VID · VizWizaccuracy: 37 -> 36.9No clear effect
use Resampler instead of Q-Former for frame selection (ablation)B-VLLM frame selection · MVBenchaccuracy: 46.5 -> 44.3Hurt
use [CLS] token features instead of mean pooling for frame selection (ablation)VideoLLaMA2 preliminary · MVBenchaccuracy: 41.2 -> 34.5Hurt

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