Inflated 3D Video Summarization: A Comprehensive Review

Video Summarization is a method that gives a short representation of the original video, which includes the summary of the key points from the entire video. Video summarization provides a clear analysis of the video by removing noise and unnecessary content and extracting key frames contents from the original video. The architecture in video summarization shows how a large video shortens to short and story contents. Many researchers studied and proposed various video summarization techniques. The proposed approaches include Convolutional Networks + Long Short-Term Memory (ConvNets+LSTM), 3 Dimensional Convolutional Networks (3D ConvNets), Two Stream Network, You Only Look Once (YOLO), Generative Adversarial Networks (GAN) and Inflated 3D ConvNet(I3D). These techniques have their own drawbacks but when compared with each other I3D gives accuracy around 93.4% for video summarization which is better as compared to others. Object detection in video summarization requires detection of spatial and temporal features. Video Summarization methods have vast applications like creating trailers of a film, generating highlights of a game, etc.