Novel TV Commercial Detection in Cookery Program Videos

In this paper, we present a simple but very effective method for detecting commercial boundaries in a cookery video program. Our approach to this problem is to first identify the best audio feature to detect commercial boundary for cookery videos with and without background music. Further, we boost the confidence level of detecting boundary using novel method of logo (program name) video frames detection indicating the start and end of the commercial at the boundaries detected by audio. Experiments show that combining audio features and video clues yields good recall and precision on a test set videos from different Indian television broadcast data.