Categorization of YouTube Videos by Video Sampling and Keyword Processing

Many students benefit from a wide variety of educational content on YouTube. It is a challenging task to watch only educational videos without any other distracting entertaining video catching our eye. Classifying the videos as educational and non-educational helps students to focus on learning without distraction. This paper describes a two-level automated mechanism to categorize videos into Educational and Non-Educational videos. The first level of classification is done by Video Processing. The feature vectors of the frames of the video are extracted using Inception V3, which are then processed using deep learning techniques. The second level of classification is done by Text Processing of the keywords available in the metadata of YouTube Videos. These keywords are used in the searching for YouTube videos. Our work brings novelty to the approach by classifying videos using both Computer Vision and Natural Language Processing techniques using the keywords from metadata. This method is faster and efficient to create a distraction-free environment where students can learn. The proposed approach has successfully classified videos into educational and non-educational.