AI Paper Atlas

Research map / Video / Video-language understanding

FrameFusion: Combining Similarity and Importance for Video Token Reduction on Large Vision Language Models

Tianyu Fu, Tengxuan Liu, Qinghao Han, Guohao Dai, Shengen Yan, Huazhong Yang, Xuefei Ning, Yu Wang

ICCV 2025 · Video › Video-language understanding · approach: Long video understanding · 49 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
compute similarity between spatially corresponding tokens of adjacent frames instead of random pairs (ablation)FrameFusion (Llava-Video-7B) · VideoNIAH/VideoMME/NExT-QA avgaverage score: 58.5 -> 64.9Improved
compute similarity between corresponding tokens of adjacent frames instead of adjacent patches (ablation)FrameFusion (Llava-Video-7B) · VideoNIAH/VideoMME/NExT-QA avgaverage score: 60.2 -> 64.9Improved
use cosine similarity instead of inner product (ablation)FrameFusion · VideoNIAH/VideoMME/NExT-QA avgaverage score: 61.7 -> 64.5Improved
use cosine similarity instead of Minkowski-1/2 distance (ablation)FrameFusion · VideoNIAH/VideoMME/NExT-QA avgaverage score: 63.1 -> 64.5Improved
merge at layer 2 instead of layer 14 (ablation)FrameFusion (Llava-Video-7B) · 50% budget avgaverage score: 50.8 -> 65.2Improved
merge at layer 2 instead of layer 13 (ablation)FrameFusion (Llava-Video-7B) · 50% budget avgaverage score: 59.2 -> 65.2Improved
merge before pruning rather than prune before merge (ablation)FrameFusion (Llava-Video-7B) · VideoNIAH/VideoMME/NExT-QA avgaverage score: 63.0 -> 63.6Improved
cascaded instead of non-cascaded merging at equal FLOPs (ablation)FrameFusion (Llava-Video-7B) · VideoNIAH/VideoMME/NExT-QA avgaverage score: 63.4 -> 65.4Improved
raise S_threshold from 0.6 to 0.9 (ablation)FrameFusion (NVILA-8B) · VideoNIAHaccuracy: 73.6 -> 72.2Mixed
raise N_threshold from 0.1 to 0.3 (ablation)FrameFusion (NVILA-8B) · VideoNIAHaccuracy: 73.3 -> 74.2No clear effect
replace FastV with FrameFusion at 30% budgetFastV (Llava-Video-7B) · 9-benchmark averageaverage score: 57.9 -> 64.8Improved
replace PruMerge with FrameFusion at 30% budgetPruMerge (NVILA-8B) · 9-benchmark averageaverage score: 52.0 -> 65.9Improved

Later papers that compared with this method

6 later accepted papers report FrameFusion as a baseline in their results tables; 5 of them report a better number on most of the results they share with it.

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