Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

Toward Accessible and Safe Live Streaming Using Distributed Content Filtering with MoQ

Andrew C. Freeman

TL;DR

This work tackles real-time content moderation for live streams, where offline analysis is insufficient due to strict latency constraints. It extends Media Over QUIC Transport (MoQ) with distributed content filtering by introducing per-user ANALYZE/FILTER workflows and an APPROVE mechanism that blocks only objectionable segments. A toy light-strobe detector demonstrates the approach, showing an end-to-end latency increase of roughly one GOP duration. The results indicate that push-based, distributed analysis can provide fine-grained safety and accessibility with acceptable delays, and the authors discuss standardization and deployment steps for real-world use.

Abstract

Live video streaming is increasingly popular on social media platforms. With the growth of live streaming comes an increased need for robust content moderation to remove dangerous, illegal, or otherwise objectionable content. Whereas video on demand distribution enables offline content analysis, live streaming imposes restrictions on latency for both analysis and distribution. In this paper, we present extensions to the in-progress Media Over QUIC Transport protocol that enable real-time content moderation in one-to-many video live streams. Importantly, our solution removes only the video segments that contain objectionable content, allowing playback resumption as soon as the stream conforms to content policies again. Content analysis tasks may be transparently distributed to arbitrary client devices. We implement and evaluate our system in the context of light strobe removal for photosensitive viewers, finding that streaming clients experience an increased latency of only one group-of-pictures duration.

Toward Accessible and Safe Live Streaming Using Distributed Content Filtering with MoQ

TL;DR

This work tackles real-time content moderation for live streams, where offline analysis is insufficient due to strict latency constraints. It extends Media Over QUIC Transport (MoQ) with distributed content filtering by introducing per-user ANALYZE/FILTER workflows and an APPROVE mechanism that blocks only objectionable segments. A toy light-strobe detector demonstrates the approach, showing an end-to-end latency increase of roughly one GOP duration. The results indicate that push-based, distributed analysis can provide fine-grained safety and accessibility with acceptable delays, and the authors discuss standardization and deployment steps for real-world use.

Abstract

Live video streaming is increasingly popular on social media platforms. With the growth of live streaming comes an increased need for robust content moderation to remove dangerous, illegal, or otherwise objectionable content. Whereas video on demand distribution enables offline content analysis, live streaming imposes restrictions on latency for both analysis and distribution. In this paper, we present extensions to the in-progress Media Over QUIC Transport protocol that enable real-time content moderation in one-to-many video live streams. Importantly, our solution removes only the video segments that contain objectionable content, allowing playback resumption as soon as the stream conforms to content policies again. Content analysis tasks may be transparently distributed to arbitrary client devices. We implement and evaluate our system in the context of light strobe removal for photosensitive viewers, finding that streaming clients experience an increased latency of only one group-of-pictures duration.
Paper Structure (19 sections, 1 equation, 2 figures)

This paper contains 19 sections, 1 equation, 2 figures.

Table of Contents

  1. Introduction
  2. Related Work
  3. HTTP Adaptive Streaming
  4. Media Over QUIC (MoQ)
  5. Content Analysis and Filtering
  6. Health Risks
  7. Illegal Content
  8. Offensive Content
  9. Motivation
  10. System Design
  11. Subscription Messages
  12. Group Approval Messages
  13. Session Management
  14. Implementation
  15. Implementation Details
  16. ...and 4 more sections

Figures (2)

  • Figure 1: Overall diagram of our distributed content filtering system. Analysis tasks can transparently be distributed to both servers (e.g., Subscriber 1) and user devices (e.g., Subscriber 2) with the same MoQ mechanisms.
  • Figure 2: Example of our strobe detection application on the infamous Pokémon Episode 38, which was reported to have caused hundreds of Japanese children to experience epileptic seizures ishida_photosensitive_1998. In our system, this flashing strobe sequence can be filtered so that photosensitive viewers are not exposed to risk.