PROCEDE DE DETECTION D'EVENEMENTS PAR VIDEOSURVEILLANCE

摘要

The process involves classifying each object into a preset category using a prediction model whose classification rules relate to object`s characteristic values. An alarm is activated if an object is classified into a category relating to an alarm situation. The model is created during an upstream learning phase where an operator indicates, in an audio/video flow sequence, in which category an object is to be present in the sequence. An independent claim is also included for a system permitting the implementation of a process for detecting predefined events.