AI Paper Atlas

Research map / Video / Video-language understanding

FLoC: Facility Location-Based Efficient Visual Token Compression for Long Video Understanding

Janghoon Cho, Jungsoo Lee, Munawar Hayat, Kyuwoong Hwang, Fatih Porikli, Sungha Choi

ICLR 2026 · Video › Video-language understanding · approach: Long video understanding · 7 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
compress with FLoC instead of DivPrune (1/32)Qwen2.5-VL-7B · Avg of 4 benchmarksaccuracy: 55.89 -> 57.45Improved
use FLoC with 7200 frames compressed vs 768 frames full tokensQwen2.5-VL-7B · Avg of Video-MME/MLVU/LVBaccuracy: 65.82 -> 67.03Improved
use FLoC with 7200 frames vs 768 frames (32B)Qwen2.5-VL-32B · Avg of Video-MME/MLVU/LVBaccuracy: 68.19 -> 70.63Improved
facility location instead of K-MeansQwen2-VL-7B · Avg accuracyaccuracy: 58.66 -> 59.74Improved
facility location instead of K-Means (compression time)Qwen2-VL-7B · 1/32 ratio T=32seconds: 59.00 -> 0.413Improved
FLoC vs Spectral Clustering on Video-MME overallLLaVA-Video-7B · Video-MME 1/8accuracy: 61.81 -> 61.04Hurt
FLoC vs Pooling on Action CountQwen2-VL-7B · MLVU AC 1/32accuracy: 32.52 -> 23.3Hurt
FLoC vs DivPrune on Needle QAQwen2-VL-7B · MLVU NQA 1/32accuracy: 68.45 -> 71.27Improved
increase block length T beyond 4 (ablation)FLoC · Qwen2-VL-7B block lengthaccuracyImproved
FLoC vs DivPrune at 1/32 on InternVL3 Video-MMEInternVL3-8B · Video-MME 1/32accuracy: 60.85 -> 60.81Hurt

Earlier methods this paper compares with

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