A semantic-based probabilistic approach for real-time video event recognition

This paper presents an approach for real-time video event recognition that combines the accuracy and descriptive capabilities of, respectively, probabilistic and semantic approaches. Based on a state-of-art knowledge representation, we define a methodology for building recognition strategies from event descriptions that consider the uncertainty of the low-level analysis. Then, we efficiently organize such strategies for performing the recognition according to the temporal characteristics of events. In particular, we use Bayesian Networks and probabilistically-extended Petri Nets for recognizing, respectively, simple and complex events. For demonstrating the proposed approach, a framework has been implemented for recognizing human–object interactions in the video monitoring domain. The experimental results show that our approach improves the event recognition performance as compared to the widely used deterministic approach.

► Semantic-based probabilistic approach for human-related event detection. ► A methodology for building recognition strategies from event descriptions. ► Uncertainty management for low-level analysis and semantic definitions. ► Tests show that the approach is able to detect five human–object interactions operating at real-time.