An Objectionable Video Classification Algorithms based on Single and Group Frame based Features

H. Lee, S. Lee, and T. Nam (Korea)


Video Classification, Machine Learning, Feature Extrac tion, Support Vector Machine, Skin Color Detection, Group of Frame


This study deals with the algorithms for objectionable video classification. The three different methods are com pared on 2000 experimental data. The categories are: (a) harmful image classification using skin color region (b) video classification using GoF (group of frame) and (c) dis criminant analysis using above two methods. This paper describes different objectionable video classification meth ods based on a feature extraction method and compares cal culation cost and accuracy of the methods. Some experi ments were carried out to find out the optimal threshold, the cost, and the performance. Our results show that the method (b) and (c) has the best performance.

Important Links:

Go Back