作者: Yuang-Tzong Lan , Sze-Yao Ni , Yi-Wei Chen , Tom C. I. Lin
DOI:
关键词:
摘要: Described herein are a system and method for abnormal behavior detection using automatic classification of multiple features. Features from various sources, including those extracted camera input through digital image analysis, used as to machine learning algorithms. These algorithms group the features produce models normal behaviors. Outlying behaviors, such identified by their lower frequency, deemed abnormal. Human supervision may optionally be employed ensure accuracy models. Once created, these can automatically classify or This invention is suitable use in traffic running red lights, driving wrong lane, against regulations.