Browse Prior Art Database

Fine-grained Video Key Frame and Clique Detection and Analytics by Synchronized Comments

IP.com Disclosure Number: IPCOM000246949D
Publication Date: 2016-Jul-18
Document File: 3 page(s) / 69K

Publishing Venue

The IP.com Prior Art Database

Abstract

This disclosure deals with three important problems: Automatic key frame/clip detection and recommendation, generation based on (current) synchronized user comments and video content. Personalized key frame/clip detection, generation and recommendation based on (historical) synchronized user comments. User preference elicitation and clustering based on synchronized user comments at clip or key frame level.

This text was extracted from a PDF file.
This is the abbreviated version, containing approximately 100% of the total text.

Page 01 of 3

Fine-grained Video Key Frame and Clique Detection and Analytics by Synchronized Comments

Time synchronized comments (bullet screen) are very popular now especially in Japan and China. Existing methods mostly use traditional visual/audio and textual information from subtitle and user comments on the video to perform analytics on the videos . This approach turn to the time synchronized comments input by the watchers when they are engaged in the video watching. This provides a natural, more robust and more fine-graied time tagging about the video frame and vide clip. This allows better vide clip and frame hilight detection, semantic analytics and personalized recommendation.

The following flowchart illustrates the key differentiation compared with existing approach.

1


Page 02 of 3

More specifically, the above techniques can be further embodied in personalized recommendation as illustrated in the following chart:
1) Automatickey frame/clip detection and generation based on
-Synchronized comments from one or multiple users

•Semantic analytics for the topic, emotion of users and its hotness etc.

2) Personalized key frame/clip detection and recommendation based on
-Synchronized comments from one or multiple users
-Find similar users by computing the common opinion on similar frames/clips, and generate the user-clip level rating matrix

2


Page 03 of 3

-Collaborative filtering to do personalized key clip generation and recommendation -Video content can be jointly used.

3