ENHANCING VIDEO RETRIVAL THROUGH BAG-OF-FEATURES AND MACHINE LEARNING FUSION

Abstract

Abstract - Content-based retrieval involves searching for information based on its content rather than qualities. The issue of content-based video retrieval (CBVR) is designing systems that can accurately and automatically process massive amounts of diverse movies. In addition, a content-based video retrieval system requires initial frame extraction. Features are derived from video frames. Finally, select an efficient similarity/classifier metric and machine learning algorithm to obtain video results linked to your query. Video frames are classified using the Random Forest Classifier, a machine learning technique, after extracting Histogram of Oriented Gradients (HOG) characteristics. Keywords: Features derived from video frames, Content Based Video Retrieval (CBVR), Histogram of Oriented Gradients, Machine Learning, Random Forest Classifier.