USER'S VISUAL INTENTION DRIVEN RETRIEVAL MODEL FOR SEMANTIC VIDEO SEARCH AND RETRIEVAL

There are various models of retrieval mechanisms for effective video retrieval, however, 'semantic gap' still persists. This study aims to reduce the gap by providing a graphical user interface that allows user to express the visuals in their mind into a query using semantic association of spatial and temporal visual concepts. To accomplish this, we have developed a video content depiction schema that serves as the foundation to our semantic search mechanisms, namely a generic visual concept search, Spatial Concept Association Based Search (SCABS) and Spatial Concept Transition Based Search (SCTBS). In our approach, a clear understanding between the user and the system has been established through the depiction schema and the visual retrieval mechanisms in contrast to common tag based retrieval systems. This retrieval model employs visual experiences and context knowledge of the users to generate search results. Our experiment result shows that our approach produces positive result against other approaches.