AI Paper Atlas

Research map / Video / Video-language understanding

VideoLLaMB: Long Streaming Video Understanding with Recurrent Memory Bridges

Yuxuan Wang, Yiqi Song, Cihang Xie, Yang Liu, Zilong Zheng

ICCV 2025 · Video › Video-language understanding · approach: Long video understanding · 46 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
replace recurrent memory with mean pooling (ablation)VideoLLaMB · EgoSchemaaccuracy: 51.61 (-2.19)Improved
add recurrent memory instead of adaptive pooling (ablation)VideoLLaMB · EgoSchemaaccuracy: 49.4 (-4.4)Improved
add memory retrieval to the memory cache (ablation)VideoLLaMB · EgoSchemaaccuracy: 52.2 (-1.6)Improved
use SceneTiling instead of uniform segmentation (ablation)VideoLLaMB · EgoSchemaaccuracy: 52 (-1.8)Improved
add image data mixture in training (ablation)VideoLLaMB · EgoSchemaaccuracy: 49.8 (-4.0)Improved
add bridge-layer output rather than memory tokens only (ablation)VideoLLaMB · EgoSchemaaccuracy: 50.4 (-3.4)Improved
use k=8 segments instead of default (ablation)VideoLLaMB · EgoSchemaaccuracy: 52.8 (-1.0)Hurt
increase memory tokens 32 to 64 with 1 bridge layer (ablation)VideoLLaMB · EgoSchemaaccuracy: 53.8 -> 53Hurt
increase bridge layers 1 to 3 and memory tokens 32 to 64 (ablation)VideoLLaMB · EgoSchemaaccuracy: 53.8 -> 54.6Improved
VideoLLaMB bridge memory instead of adaptive pooling compression (ablation)VideoLLaMB · EgoSchemaaccuracy: 45.6 -> 53.8Improved
train on VideoChat2 data instead of PLLaVA dataVideoLLaMB · MVBenchaverage accuracy: 49.33 -> 52.5Improved
dynamic segments instead of static 4 segments for length extrapolation (ablation)VideoLLaMB · EgoSchemaaccuracyMixed

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