|
|
Research and Development
In this project, we shall use semantic videos analysis techniques to generate semantic video description. The description will then be used to match against user input conceptual query.
|
Technologies to be developed
|
- Semantic Video Analysis
This technology is related to the generation of the semantic descriptions from the second-level video analysis, which will be used to match against the conceptual query specified by the users. In this part, more detail analysis will be performed on the candidate videos obtained from the first-level retrieval process. It may need to perform object recognitions such as [28-30,35-38] (see 3.2.1) (e.g. differentiate a human from a vehicle) and derive interactions among the objects [7,36,37] (e.g. The man is getting into the vehicle) so that the queries ¡§The man is getting into the vehicle¡¨ and ¡§The vehicle is hitting the man¡¨ will generate different but relevant retrieval results.
Innovative use of existing technologies
|
-
This project will also use existing technologies in semantic video analysis innovatively to perform accurate semantic descriptor generation for the second-level processing in the hierarchical video retrieval system. The research work involved in this aspect includes extensive study and trial of large combinations of different video analysis tools for achieving the best performance.
|
|
|
|
|
|
