Incremental Learning Approach for Events Detection from Large Video Dataset

In this paper, we propose a strategy of multi-SVM incremental learning system based on Learn++ classifier for detection of predefined events in the video. This strategy is offline and fast in the sense that any new class of event can be learned by the system from very few examples. The extraction and synthesis of suitably video events are used for this purpose. The results showed that the performance of our system is improving gradually and progressively as we increase the number of such learning for each event. We then demonstrate the usefulness of the toolbox in the context of feature extraction, concepts/events learning and detection in large collection of video surveillance dataset.