AUTOMATIC ONLINE LECTURE HIGHLIGHTING BASED ON MULTIMEDIA ANALYSIS
E. NEHA REDDY, G. RAHUL KUMAR, M. VIJAYA(ASSISTANT
PROFESSOR), B. DEEKSHIKA, G. NAVEEN PAVAN SAI
Dept. Of Computer Science And Engineering, Vidya Jyothi Institute Of Technology, Hyderabad
ABSTRACT:
The benefits of textbook highlighting for pupils are generally acknowledged. In this essay, we offer a thorough solution for highlighting the online lectures at the segment and sentence levels, much like it is done with printed books. this age of e-learning, especially with MOOCs, the answer is built on the automatic analysis of multimedia lecture materials like speeches, transcripts, and slides. Sentence-level lecture highlighting makes use of acoustic information from the audio and is incorporated in subtitle files of associated MOOC videos. The precision is over 60% when compared to expert-created ground truth, which is higher than baseline results and well-liked by users. While using statistical analysis, segment-level lecture highlighting primarily examines the speech transcripts, the lecture slides, and their linkages. A review method reveals that general accuracy can reach 70% with the ground truth produced by large numbers of users, which is pretty encouraging. Key-frame detection and automated video segmentation are used. Next, it extracts/takes out textual meta-data by using video Optical Character Recognition (OCR) technology on lecture video key-frames and Automatic Speech Recognition (ASR) on lecture audio tracks content to get audio from the video and then convert that audio track into text information. The primary goal is to provide extracted words with multimedia data, such as photos and movies. Online education is no more a sophisticated idea reserved for the ivory tower in this day of high-speed globalisation and information technology; rather, it is a fast growing sector with direct relevance to people's everyday lives.
keywords: MOOC, OCR, ASR, Video Segmentation, Key
frame detection, Meta Data
