Cricket Videos Summary generation using a Novel Convolutional Neural Network

Abstract

Video is the combination of frames, many sequences of images known as frames are grouped together to form a video. The exponential growth of video data requires constant exploration from original videos by extracting only informative content present in the video. Traditional video processing applications process frame one by one and consume a lot of time, however, video summarization techniques based on deep learning extract only key-frames in the video based on classification for informative content. Supervised and unsupervised video processing techniques help people to reduce their efforts for video summarization. In this paper, we propose a novel customized convolutional neural network i.e. a supervised model of deep learning for summarization of cricket videos. Our proposed Cricket- Convolutional Neural Network (C-CNN) learns the most informative features from video frames and performs binary classification into positive and negative class. We have performed an extensive experimentation which ensures that our proposed C-CNN network outperforms the existing techniques for cricket video summarization.