Distinguishing Excitement Movie Clips using Movement and Pace Features

S.C. Watanapa, B. Thipakorn, and N. Charoenkitkarn (Thailand)


Emotion-based Classification, Image Processing and Analysis, Semantic Classification, and Video Content Analysis


Classifying multimedia data into semantic, such as emotional, categories is challenging and useful. Using two film artistic components: movement and pace, this paper proposes a method that can distinguish movie clips belonging to the excitement class from the others. Movement is represented by the Average Squared Motion Vector Magnitude feature, and pace is represented by the Average Shot Duration feature. Classification experiments with 101 data clips, excerpted from 24 Hollywood movies, are conducted, employing minimum distance and k-NN classifiers. The results show that the selected features can potentially separate the excitement movie clips from the others with above 90% accuracy.

Important Links:

Go Back