A Review on Video to Text Summarization Techniques

Abstract

In the recent years of technological developments in virtually each sector be it CCTV, Social Media Platforms, Medical, Defense etc., a large amount of video data is created every day. The processing of such large chunky videos necessitates a lot of storage, lot of computational processing power, as well lots of time. Furthermore, some videos are long and boring, or they take an enormous amount of time to display information that can be found in a matter of minutes. As an analyst need to watch the entire video, extracting features is a time-consuming task. There are numerous editing tools available, but they all require expertise. Video summarization is a technique used to overcome the problems associated with long videos. This technique condenses the videos based on the various features. Text summarizing automatically provides a summary that covers all pertinent key information from the original video, which can be divided into two main approaches: extractive summarization and abstractive summarization. The extractive summarization is maturing and research is now shifting towards abstractive summation. Nowadays, several deep learning techniques like CNN, RNN, LSTM, and clustering algorithms such as hierarchical KNN are used for the abstractive text summarization. In this review work, we examined all the strategies utilized for text summarizing during the last ten years on various parameters.