Video Annotation Methodology Based on Ontology for Transportation Domain

Increase in the importance of video data has led to flooding of video content over the internet and offline world. For such abundance of content the access of relevant video in reduced time is a matter of concern. So the need for a system which will assist in content based access of video arises. This paper introduces a novel approach for video annotation. The key frames are extracted from the video and are analyzed. Instead of complete video frames, only the key frames are analyzed to identify the objects present. The object detectors are trained for identification of the object. The detected objects are then added in the annotation file. The annotation is based on ontology which eases the semantic retrieval of videos. Ontology based video annotation greatly accelerates the performance of retrieval systems.