A method for video categorization by analyzing text, audio, and frames
- PDF / 797,149 Bytes
- 10 Pages / 595.276 x 790.866 pts Page_size
- 5 Downloads / 199 Views
ORIGINAL RESEARCH
A method for video categorization by analyzing text, audio, and frames Hossain Md Al Amin1 • Mohammad Shamsul Arefin1 • Pranab Kumar Dhar1
Received: 28 November 2018 / Accepted: 29 July 2019 Bharati Vidyapeeth’s Institute of Computer Applications and Management 2019
Abstract A video file naturally contains audio, text metadata, and visual content in the form of frames, as it is a series of images with adequate motion. To get an efficient result in video categorization, it is necessary to use and analyze all the available resources. For this reason, in this paper we introduce a video categorization method by examining all the essential elements of video in the form of text, audio, and frames. The proposed method consists of three different modules. These modules are used for analyzing the text, audio, and visual contents to provide the analysis results, which are finally combined to get the final output. A set of fundamental properties are analyzed and compared with standard values acquired from training data set to understand the genre of the videos and eventually tagging it with the most probable category. Besides, we have conducted different tests using the proposed method and the simulation results show that the proposed method effectively categorizes the video sequence. Keywords Video categorization Text mining Supervised learning Image processing Signal analysis
& Mohammad Shamsul Arefin [email protected] Hossain Md Al Amin [email protected] Pranab Kumar Dhar [email protected] 1
Department of Computer Science and Engineering, Chittagong University of Engineering and Technology (CUET), Chittagong-4349, Bangladesh
1 Introduction A video is the combination of still images, which are called frames and in most of the case contains audio that correlates with the visual contents. It is a strong and immensely popular medium of entertainment as well as education or communication. According to the DMR [1], 1 billion hours of videos are watched on YouTube per day, which is considered as the most popular video content platform so far. People spent 40 min on an average basis on YouTube per day. These reflect on how important it is in our regular life. With the modern technology and as a prime form of entertainment, we need to deal with an enormous number of videos. Benefits of having the significant amount of videos would be hugely reduced if these online videos are not correctly classified and tagged. Also, to provide the right contents to the interested group of people, video classification needs to be drawn into consideration, as the automatic labeling of video footage is a standard requirement to deal with indexing of extensive collection of videos. Because of the semantic differences between a text and a video file, the automatic video categorization problem is far different from the other types of document classification. A video file contains audio, video and also text where other document files only include the text dimension. For this reason, it is not easy to develop a
Data Loading...
