Enregistré dans:
Détails bibliographiques
Auteurs principaux: Wang, Han, Ji, Deyi, Zhu, Lanyun, Luo, Jiebo, Lee, Roy Ka-Wei
Format: Preprint
Publié: 2026
Sujets:
Accès en ligne:https://arxiv.org/abs/2601.22738
Tags: Ajouter un tag
Pas de tags, Soyez le premier à ajouter un tag!
_version_ 1866908799913164800
author Wang, Han
Ji, Deyi
Zhu, Lanyun
Luo, Jiebo
Lee, Roy Ka-Wei
author_facet Wang, Han
Ji, Deyi
Zhu, Lanyun
Luo, Jiebo
Lee, Roy Ka-Wei
contents Live streaming platforms require real-time monitoring and reaction to social signals, utilizing partial and asynchronous evidence from video, text, and audio. We propose StreamSense, a streaming detector that couples a lightweight streaming encoder with selective routing to a Vision-Language Model (VLM) expert. StreamSense handles most timestamps with the lightweight streaming encoder, escalates hard/ambiguous cases to the VLM, and defers decisions when context is insufficient. The encoder is trained using (i) a cross-modal contrastive term to align visual/audio cues with textual signals, and (ii) an IoU-weighted loss that down-weights poorly overlapping target segments, mitigating label interference across segment boundaries. We evaluate StreamSense on multiple social streaming detection tasks (e.g., sentiment classification and hate content moderation), and the results show that StreamSense achieves higher accuracy than VLM-only streaming while only occasionally invoking the VLM, thereby reducing average latency and compute. Our results indicate that selective escalation and deferral are effective primitives for understanding streaming social tasks. Code is publicly available on GitHub.
format Preprint
id arxiv_https___arxiv_org_abs_2601_22738
institution arXiv
publishDate 2026
record_format arxiv
spellingShingle StreamSense: Streaming Social Task Detection with Selective Vision-Language Model Routing
Wang, Han
Ji, Deyi
Zhu, Lanyun
Luo, Jiebo
Lee, Roy Ka-Wei
Computer Vision and Pattern Recognition
I.4
Live streaming platforms require real-time monitoring and reaction to social signals, utilizing partial and asynchronous evidence from video, text, and audio. We propose StreamSense, a streaming detector that couples a lightweight streaming encoder with selective routing to a Vision-Language Model (VLM) expert. StreamSense handles most timestamps with the lightweight streaming encoder, escalates hard/ambiguous cases to the VLM, and defers decisions when context is insufficient. The encoder is trained using (i) a cross-modal contrastive term to align visual/audio cues with textual signals, and (ii) an IoU-weighted loss that down-weights poorly overlapping target segments, mitigating label interference across segment boundaries. We evaluate StreamSense on multiple social streaming detection tasks (e.g., sentiment classification and hate content moderation), and the results show that StreamSense achieves higher accuracy than VLM-only streaming while only occasionally invoking the VLM, thereby reducing average latency and compute. Our results indicate that selective escalation and deferral are effective primitives for understanding streaming social tasks. Code is publicly available on GitHub.
title StreamSense: Streaming Social Task Detection with Selective Vision-Language Model Routing
topic Computer Vision and Pattern Recognition
I.4
url https://arxiv.org/abs/2601.22738