Saved in:
Bibliographic Details
Main Authors: Chen, Jiahui, Peng, Bo, Jia, Lianchen, Zhang, Zeyu, Huang, Tianchi, Sun, Lifeng
Format: Preprint
Published: 2026
Subjects:
Online Access:https://arxiv.org/abs/2602.14214
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866911449758040064
author Chen, Jiahui
Peng, Bo
Jia, Lianchen
Zhang, Zeyu
Huang, Tianchi
Sun, Lifeng
author_facet Chen, Jiahui
Peng, Bo
Jia, Lianchen
Zhang, Zeyu
Huang, Tianchi
Sun, Lifeng
contents Content-aware streaming requires dynamic, chunk-level importance weights to optimize subjective quality of experience (QoE). However, direct human annotation is prohibitively expensive while vision-saliency models generalize poorly. We introduce HiVid, the first framework to leverage Large Language Models (LLMs) as a scalable human proxy to generate high-fidelity weights for both Video-on-Demand (VOD) and live streaming. We address 3 non-trivial challenges: (1) To extend LLMs' limited modality and circumvent token limits, we propose a perception module to assess frames in a local context window, autoregressively building a coherent understanding of the video. (2) For VOD with rating inconsistency across local windows, we propose a ranking module to perform global re-ranking with a novel LLM-guided merge-sort algorithm. (3) For live streaming which requires low-latency, online inference without future knowledge, we propose a prediction module to predict future weights with a multi-modal time series model, which comprises a content-aware attention and adaptive horizon to accommodate asynchronous LLM inference. Extensive experiments show HiVid improves weight prediction accuracy by up to 11.5\% for VOD and 26\% for live streaming over SOTA baselines. Real-world user study validates HiVid boosts streaming QoE correlation by 14.7\%.
format Preprint
id arxiv_https___arxiv_org_abs_2602_14214
institution arXiv
publishDate 2026
record_format arxiv
spellingShingle HiVid: LLM-Guided Video Saliency For Content-Aware VOD And Live Streaming
Chen, Jiahui
Peng, Bo
Jia, Lianchen
Zhang, Zeyu
Huang, Tianchi
Sun, Lifeng
Computer Vision and Pattern Recognition
Content-aware streaming requires dynamic, chunk-level importance weights to optimize subjective quality of experience (QoE). However, direct human annotation is prohibitively expensive while vision-saliency models generalize poorly. We introduce HiVid, the first framework to leverage Large Language Models (LLMs) as a scalable human proxy to generate high-fidelity weights for both Video-on-Demand (VOD) and live streaming. We address 3 non-trivial challenges: (1) To extend LLMs' limited modality and circumvent token limits, we propose a perception module to assess frames in a local context window, autoregressively building a coherent understanding of the video. (2) For VOD with rating inconsistency across local windows, we propose a ranking module to perform global re-ranking with a novel LLM-guided merge-sort algorithm. (3) For live streaming which requires low-latency, online inference without future knowledge, we propose a prediction module to predict future weights with a multi-modal time series model, which comprises a content-aware attention and adaptive horizon to accommodate asynchronous LLM inference. Extensive experiments show HiVid improves weight prediction accuracy by up to 11.5\% for VOD and 26\% for live streaming over SOTA baselines. Real-world user study validates HiVid boosts streaming QoE correlation by 14.7\%.
title HiVid: LLM-Guided Video Saliency For Content-Aware VOD And Live Streaming
topic Computer Vision and Pattern Recognition
url https://arxiv.org/abs/2602.14214