Location: Zoom
Automated Analysis and Indexing of Lecture Videos Using OCR and ASR Technology.
Many online educational organizations and universities use video lectures to support teaching and learning. However, it is difficult and time-consuming for students to search for information within these videos. One observation is that the text displayed in these video lectures is a potential resource for analyzing and indexing the lecture contents. We present an approach for automatic analysis and indexing of lecture videos using a combination of OCR (Optical Character Recognition) and ASR (Automatic Speech Recognition) technologies. To accomplish this task, we search for and identify frames with substantial text content from a lecture video and then use OCR to identify the textual contents. In addition, we also use ASR to augment and validate this information. The entire video is then indexed based on the occurrence of text and meta-data is added at the beginning of these segments. We have built a prototype to demonstrate our approach to make searching for information within lecture videos feasible.
Committee: Simanta Mitra (major professor), Gurpur Prabhu (major professor), and Carl Chang