J. Bannour, B. Georis, F. Brémond, and M. Thonnat (France)
3D animation, scenario models, evaluation of video in terpretation algorithms, video interpretation, synthetic movies.
The objective of this work is to generate automatically 3D animations from scenario models defined by human experts for scenario recognition. These animations are useful for two reasons. First, they can help experts to validate their scenario models. Second, synthetic video sequences can be generated from these 3D animations and serve as refer ence data to evaluate video interpretation algorithms. An animation is generated in two stages. First, we transform a scenario model into a visualisation model. Second, we gen erate animation instances from visualisation models. The first results obtained are promising. We have automatically generated 3D animations for two applications: the moni toring of a bank agency and a metro platform.
Important Links:
Go Back
