AI Paper Atlas

Research map / Video / Video-language understanding

HiVid: LLM-Guided Video Saliency For Content-Aware VOD And Live Streaming

Jiahui Chen, Bo Peng, Lianchen Jia, Zeyu Zhang, Tianchi Huang, Lifeng Sun

ICLR 2026 · Video › Video-language understanding · approach: Long video understanding · 1 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 LLM-guided ranking module (ablation)HiVid · Youtube-8M saliencyPLCC: 0.611 -> 0.660Improved
add perception summary module (ablation)HiVid · Youtube-8M saliencyPLCC: 0.632 -> 0.660Improved
add Gaussian smoothing (ablation)HiVid · Youtube-8M saliencyPLCC: 0.619 -> 0.660Improved
add correlation loss to MSE (ablation)HiVid-M · time series forecastingPLCC: 0.21 -> 0.29Improved
use adaptive output length instead of constant L_out (ablation)HiVid-M · live overall PLCC m=10PLCC: 0.10/0.17/0.18 -> 0.20Improved
use GPT-4o instead of open-source MLLMs (ablation)HiVid · Youtube-8M saliencyPLCC: 0.58-0.64 -> 0.66Improved
use LLM prompting for forecasting instead of trained modelLLM forecaster · time series forecastingRMSE: HiVid-M 0.12 -> LLM 0.27Hurt
use more complex forecasting architecturesiTransformer/TimesNet etc. · time series forecastingPLCC: HiVid-U 0.24 vs Crossformer 0.23No clear effect
increase window length m from 2 to 10 (ablation)HiVid · VOD overheadtotal cost per video ($): 8.12 -> 1.35Improved
increase window length m (ablation)HiVid live · live utilityforecasting utilityHurt
use GPT-4o-mini with m=2 instead of GPT-4o m=10 (ablation)HiVid · VOD saliencyPLCC: 0.660 -> 0.645; cost 1.35 -> 0.44Hurt

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