AI Paper Atlas

Research map / Video / Video-language understanding

Accelerating Streaming Video Large Language Models via Hierarchical Token Compression

Yiyu Wang, Xuyang Liu, Xiyan Gui, Xinying Lin, Boxue Yang, Chenfei Liao, Tailai Chen, Linfeng Zhang

CVPR 2026 · Video › Video-language understanding · approach: Video LLMs with fine-grained understanding · 31 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 STC-Pruner (75% prune) instead of ToMe/VisionZipReKV · OVO-Bench overallaccuracy: 46.4 -> 50.6Improved
add STC-Pruner vs VidCom2ReKV · StreamingBench overallaccuracy: 63.6 -> 63.7No clear effect
add STC-Cacher & Pruner jointlyReKV · OVO-Bench overallaccuracy: 52.6 -> 52.0Hurt
add STC-Cacher & Pruner jointly (30% tokens)ReKV · StreamingBench overallaccuracy: 69.1 -> 65.2Hurt
add STC-Cacher & Pruner jointlyReKV · OVO-BenchViT encoding latency (s): 103.7 -> 78.3Improved
add STC-Cacher & Pruner jointlyReKV · OVO-BenchLLM prefilling latency (s): 482.4 -> 263.7Improved
add STC-Cacher to ViT of streaming modelsStreamForest · OVO-Bench overallaccuracy: 54.3 -> 52.3Hurt
add STC-CacherStreamForest · StreamingBench overallaccuracy: 77.3 -> 76.9No clear effect
add STC-CacherLiveCC · OVO-Bench overallaccuracy: 59.7 -> 57.3Hurt
reuse only attention features in STC-Cacher (ablation)STC-Cacher · EgoSchemaaccuracy: 59.0 -> 26.2Hurt
reuse only MLP features in STC-Cacher (ablation)STC-Cacher · EgoSchemaaccuracy: 59.0 -> 57.1Hurt
use both spatial and temporal anchors in STC-Pruner (ablation)STC-Pruner · OVO-Bench STUaccuracy: 47.2 -> 48.9Improved

Earlier methods this paper compares with

Named baselines from earlier accepted papers in this paper's results tables: better than 1 of 2 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