Abstract: Semantic search based video retrieval is hard problem due to limited set of vocabulary. Textual queries semantic search based video retrieval is able to detect object from video frames, first the sentence is converted into parse tree than noun, verb, adverb present in that sentence is converted into semantic graph. And construct the semantic meaningful graph gives the semantic structure and matches the nouns, verb and adverb detected in the video frame and also detect the action and position of the object by using semantic meaningful graph. The developed approach is to first sentence is converted into parse tree and object detection takes place. First textual query is matched to concept. The advantage of textual queries approach is object appearance and motion by using structure prediction. Textual queries semantic search can contain temporal and spatial information about multiple objects like trees and building present in the scene.

Keywords: Parse tree, Textual queries, Semantic graph, Semantic structure, Structure prediction.