【CCF-AIR青年基金】视频来源取证技术研究 Video Origin Identification over Online Social Networks

Research Themes

Other

Background

In recent years, with the upgrading of basic information technologies such as network technology, storage technology, coding technology, video content has shown a rapid development trend on the Internet, and has gradually become one of the main information transmission methods for the public. Video-related products of various lengths and forms have shown a surge. We can see that the short and medium video platforms represented by Tictok, Kuaishou, Bilibili, etc. have sprung up suddenly, with explosive growth from user volume to content volume; we can see that various online video live broadcasts are gradually becoming more and more popular among the public, and has begun to affect almost all industries; and we also can see that traditional video content such as movies and TV series has begun to be deeply Internet-based, online video has begun to compete with traditional film and television media for market share.

 

From the user's point of view, on the one hand, the development of software and hardware equipment, video platforms, and network technology has greatly reduced the cost of making and sharing videos, and each user can easily become a producer of video content. This has also resulted in a huge number of videos of various types spread to almost all corners of the Internet that provide video services. On the other hand, for video content audiences, it is more and more convenient to watch videos, with high degree of freedom and immediacy of video content selection. Compared with static media content such as images and text, video can convey information more comprehensively and intuitively. Video content is also easier for the public to understand. It can be seen that video content is gradually becoming one of the main multimedia content of the Internet, and its proportion is increasing day by day, with huge potential and development space.

 

From the perspective of Alibaba Group, video content has already involved all aspects of our business. The big entertainment represented by Youku and Alibaba Pictures is bound to focus on video content; Taobao live broadcasts and product introduction video on Taobao make the correlation between video content and Alibaba's core e-commerce business more and more closely. It is foreseeable that  a massive number of product introductions will be presented in the form of videos in the future. Moreover, in other scenarios of Alibaba Group, the proportion of video content is also increasing. It is undeniable that video content is also becoming a pivotal traffic portal in Alibaba and is becoming more and more important.

 

With the popular of online video content, a lot of related security issues are gradually exposed, especially when the source of video content is uploaded by normal users. For example, content security issues related to pornography, terrorism, violence, and politics; copyright issues such as piracy, plagiarism, and secondary editing; Traffic security issues such as on-hook, tampering, pseudo video streaming, etc. These new security issues have brought many new troubles to business operations and platform governance, and have also tested the Alibaba's technology accumulation in the video field, urging a new round of technology development and iteration.

 

However, it is difficult for us to solve all the problems mentioned above through a set of solutions. In this proposal, we focus on the security issues caused by video origin, including but not limited to piracy issue, recording and broadcasting phenomenon, tampering operation, and fake video streams involved in online video sources. We plan to develop an efficient forensics technology to identify the origin platform of user-uploaded image, and provide corresponding solutions for related business scenarios.

Target

  • Data set construction: An efficient video crawler to obtain video data from online platforms, A video data set with a large amount of online and off-line data.
  • A feasible solution or method for video content analysis and feature extraction for target platforms.
  • A feasible and effective video origin identification algorithm.


Related Research Topics

  • Image Forensics/Anti-Forensics: Image origin/source identification; Image manipulation chain detection; Image/video operation trace removal.
  • Multimedia Compression: Double/multiple multimedia compression detection; Multimedia compression parameter analysis.
  • Video Coding: Online Social Networks
  • Pattern Recognition and Artificial intelligence: Logo Recognition; Multimedia content captioning; Multimedia semantic segmentation;

Scan QR code
关注Ali TechnologyWechat Account