||• Emergency/Event Detection
- Our goal is to develop an algorithm and software architecture for detecting emergency or anomalous events in images or videos via training large-scale data-sets.
||The large-scale emergency/event detection system that responds immediately to designated situations can be applied to various areas. For maximizing its advantage of real-time event detection, the research includes optimization of detection accuracy and improvement of computational efficiency. At first, we research on applications of deep learning architectures(CNN) for processing millions of data, used in ImageNet competition. Based on the results from the preceding research, we develop algorithms/architectures processing billions of data.
||By detecting emergency or anomalous events in images or videos in real-time, various events taken from extensive areas can be recognized and responded immediately with lower cost compared with human surveillance. The intelligent video analytics can be applied to various areas such as automatic tracking, antitheft system, crackdown on illegal parking, etc.