Video Summarization using Convolutional Neural Network and Random Forest Classifier

Video summarization methods aim to generate a shortened representation of the original video. A novel method to extract key-frames based on Convolutional Neural Network and Random Forest Classifier is presented in this paper. The method processes videos on frame by frame basis. The redundant frames are first eliminated based on displacement vectors between the consecutive frames. The high-level feature vectors are extracted using CNN. The feature descriptors corresponding to frames are further classified into key-frames and non-keyframes using the Random Forest Classifier. The method is tested on two benchmark datasets: VSUMM and OVP. The proposed approach attains better results compared to other state-of-the-art video summarization techniques. The results show that the method is able to generate high quality summaries consistently for videos of all categories.