AI Paper Atlas

Research map / Video / Video-language understanding

Flash-VStream: Efficient Real-Time Understanding for Long Video Streams

Haoji Zhang, Yiqin Wang, Yansong Tang, Yong Liu, Jiashi Feng, Xiaojie Jin

ICCV 2025 · Video › Video-language understanding · approach: Video LLMs with fine-grained understanding · 43 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
Use Flash Memory (CSM + DAM) instead of Qwen2-VL under the same 11520 token budgetQwen2-VL-7B · EgoSchemaaccuracy (%): 64.0 -> 68.2Improved
Use Flash-VStream on Video-MME long videos with subtitles versus Qwen2-VL-onlineQwen2-VL-7B · Video-MME (w/ subs), Longaccuracy (%): 57.9 -> 61.4Improved
Use Flash-VStream (11520 tokens) versus Qwen2-VL full (24576 tokens) on MLVUQwen2-VL-7B · MLVU devaccuracy (%): 66.0 -> 66.3Improved
Replace uniform sampling with K-means clustering memory (CSM) (ablation)Flash Memory (CSM only) · EgoSchemaaccuracy (%): 63.4 -> 66.8Improved
Add feature-centric DAM instead of uniform-sampled DAM (ablation)Flash Memory · avg of MVBench, EgoSchema, Video-MMEaverage accuracy (%): 64.2 -> 64.9Improved
Add CSM clustering memory instead of uniform sampling (ablation)Flash Memory · avg of MVBench, EgoSchema, Video-MMEaverage accuracy (%): 61.6 -> 63.6Improved
Use Euclidean distance instead of cosine similarity in DAM retrieval (ablation)DAM · EgoSchemaaccuracy (%): 66.5 -> 68.2Improved
Select top-k largest clusters instead of top-k smallest for DAM (ablation)DAM · Video-MME (w/o subs)accuracy (%): 59.6 -> 61.2Improved
Feature-centric retrieval instead of temporal-centric retrieval (ablation)DAM · MVBenchaccuracy (%): 64.4 -> 65.4Improved
Use DBScan instead of K-means for CSM (ablation)CSM · EgoSchemaaccuracy (%): 66.8 -> 66.6No clear effect
Increase DAM from 30 to 60 key frames (11520 to 19200 tokens) (ablation)Flash Memory · avg of EgoSchema, MVBench, Video-MMEaverage accuracy (%): 64.9 -> 65.1No clear effect
Reduce CSM size from 60 to 15 clusters (ablation)Flash Memory · avg of EgoSchema, MVBench, Video-MMEaverage accuracy (%): 63.6 -> 61.5Hurt

Later papers that compared with this method

12 later accepted papers report Flash-VStream as a baseline in their results tables; 12 of them report a better number on most of the results they share with it.

4 more not listed.

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