Retrieving Instructional Video Content from Speech and Text Information

The interest of today’s generation to learn from video lectures is becoming popular due to its considerable advantages and easy availability than classroom learning. To involve into this, many institutes and organizations are using this method for teaching and learning. An enormous amount of data is generated in video lecturing form. To extract the desired information from the desired video from this vast video information available on internet becomes difficult. In this paper, we have used techniques for automatically retrieving the information from video files to collect it as a metadata for those files. For efficient retrieval of text from videos we use the OCR (Optical Character Recognition) tool to extract text from slides and ASR (Automatic Speech Recognition) tool for recognizing information from speech given by the speaker. First, we do segmentation and classification of video frames for identifying the key frames. Then the OCR and ASR tool is used for extracting the information from video slides and audio speech respectively. The collected data can be stored as a metadata for the file. Finally, the search can be made more efficient by applying clustering and ontology concept.