Action recognition based on semantic feature description and cross classification

Action recognition is a challenging topic in computer vision. In this work, we present a novel method for action recognition which is based on two claimed contributions: semantic feature description and cross classification. The designed descriptor is combined by several local 3D-SIFT and is informative and distinctive, reflecting the spatio-temporal clues of the video. The cross classification effectively combines the feature localization and action categorization together. The proposed method is justified on a popular dateset named UCF50 and the experimental results demonstrate that our method outperforms the state-of-the-art competitors.