METHOD FOR AUDIO-VISUAL EVENTS CLASSIFICATION AND LOCALIZATION AND CORRESPONDING APPARATUS COMPUTER READABLE PROGRAM PRODUCT AND COMPUTER READABLE STORAGE MEDIUM

摘要

A method and an apparatus for audio-visual events classification and localization in a video sequence are disclosed. A set of audio event proposals from an audio stream of the video sequence and a set of visual object proposals from images of the video sequence are obtained. A first score for the set of visual object proposals is determined with regards to a set of audio-visual event classes, A second score for the set of audio event proposals is determined with regards to the set of audio-visual event classes. A video-level score is determined by combining the first score and the second score.