The Effects of Short Video-Sharing Services on Video Copy Detection

TL;DR

The effects of short video-sharing services on VCD are examined by constructing a dataset that has short video-sharing service characteristics and it is seen that segment-level VCD in short video-sharing services is more difficult than those in general video-sharing services and video alignment component mainly suppress the detection performance in short video-sharing services.

Abstract

The short video-sharing services that allow users to post 10-30 second videos (e.g., YouTube Shorts and TikTok) have attracted a lot of attention in recent years. However, conventional video copy detection (VCD) methods mainly focus on general video-sharing services (e.g., YouTube and Bilibili), and the effects of short video-sharing services on video copy detection are still unclear. Considering that illegally copied videos in short video-sharing services have service-distinctive characteristics, especially in those time lengths, the pros and cons of VCD in those services are required to be analyzed. In this paper, we examine the effects of short video-sharing services on VCD by constructing a dataset that has short video-sharing service characteristics. Our novel dataset is automatically constructed from the publicly available dataset to have reference videos and fixed short-time-length query videos, and such automation procedures assure the reproducibility and data privacy preservation of this paper. From the experimental results focusing on segment-level and video-level situations, we can see that three effects: "Segment-level VCD in short video-sharing services is more difficult than those in general video-sharing services", "Video-level VCD in short video-sharing services is easier than those in general video-sharing services", "The video alignment component mainly suppress the detection performance in short video-sharing services".

Figures (6)

• Figure 1: The brief overview of the VCD in this paper. In VCD, query video is created from the part of the original reference videos by applying various edits. By detecting the query video and copy region based on the reference videos, illegally reproduced videos can be accurately detected for realizing safe and reliable short video-sharing services.
• Figure 2: The VCSL
• Figure 3: Our Dataset
• Figure 5: The overview of the dataset construction process.
• Figure 6: The time distribution of reference and query videos in VCSL dataset and our dataset ($t=10$). Different from the VCSL dataset which includes widely distributed query videos, our dataset only includes fixed-time query videos.