Abstract:Video summarization aims to analyze the structure and content of videos and extract key segments to construct summarization that can accurately summarize the main content, allowing users to quickly access the core information without browsing the full video. However, existing methods have difficulties in capturing long-term dependencies when dealing with long videos. On the other hand, there is a large amount of noise in graph structures, which may lead to the influence of redundant information and is not conducive to the effective learning of video features. To solve the above problems, we propose a video summarization generation network based on dynamic graph contrastive learning and feature fusion, which mainly consists of three modules: feature extraction, video encoder, and feature fusion. Firstly, we compute the shot features and construct a dynamic graph by using the shot features as nodes of the graph and the similarity between the shot features as the weights of the edges. In the video encoder, we extract the temporal and structural features in the video using stacked L-G Blocks, where the L-G Block consists of a bidirectional long short-term memory network and a graph convolutional network. Then, the shallow-level features are obtained after processing by L-G Blocks. In order to remove the redundant information in the graph, graph contrastive learning is used to obtain the optimized deep-level features. Finally, to fully exploit the feature information of the video, a feature fusion gate using the gating mechanism is designed to fully fuse the shallow-level features with the deep-level features. Extensive experiments are conducted on two benchmark datasets, TVSum and SumMe, and the experimental results show that our proposed method outperforms most of the current state-of-the-art video summarization methods.

Category Driven Deep Recurrent Neural Network for Video Summarization

Learning User Interest with Improved Triplet Deep Ranking and Web-Image Priors for Topic-Related Video Summarization.

Creating Personalized Video Summaries Via Semantic Event Detection

Creating Memorable Video Summaries That Satisfy the User's Intention for Taking the Videos.

A GAN Based Video Summarization Method with Representation Loss

An Unsupervised Video Summarization Method Based on Multimodal Representation.

Automatic Document Summarization Via Deep Neural Networks

From Thumbnails to Summaries - A single Deep Neural Network to Rule Them All

Video Summarization Generation Model Based on Transformer and Deep Reinforcement Learning

Video summarization via knowledge-aware multimodal deep networks

Deep Semantic and Attentive Network for Unsupervised Video Summarization

Video Summarization Generation Network Based on Dynamic Graph Contrastive Learning and Feature Fusion

Video Summarization Using Deep Neural Networks: A Survey

TTH-RNN: Tensor-Train Hierarchical Recurrent Neural Network for Video Summarization

Video Summarisation by Classification with Deep Reinforcement Learning

Action Parsing-Driven Video Summarization Based on Reinforcement Learning

Video Summarization with Long Short-term Memory

Video Summarization by Learning Deep Side Semantic Embedding

Efficient video summarization through MobileNetSSD: a robust deep learning-based framework for efficient video summarization focused on objects of interest

Reconstructive Sequence-Graph Network for Video Summarization

Dynamic graph convolutional network for multi-video summarization