Two-stage anchor-based dynamic video abstraction method

A video summarization and stage technology, applied in the field of video summarization, can solve the problems that LSTM cannot run in parallel, the network cannot run in parallel, and the network implementation is complicated, so as to reduce the training time, reduce the difficulty of optimization, and simplify the network structure.

CN113505266AActive Publication Date: 2021-10-15NANJING UNIV OF POSTS & TELECOMM
4 Cites 0 Cited by

Patent Information

Authority / Receiving Office
CN · China
Patent Type
Applications(China)
Current Assignee / Owner
Publication Date
2021-10-15

Smart Images

  • Figure 1
    Figure 1
  • Figure 2
    Figure 2
  • Figure 3
    Figure 3
Patent Text Reader

Abstract

The invention relates to a two-stage anchor-based dynamic video abstraction method. The method comprises the following steps: (1) constructing an anchor-based neural network, wherein the neural network comprises an extraction network, a self-attention mechanism network, a two-stage network, namely a region recommendation network and a regression classification network; and (2) determining types and lengths of anchors in the region recommendation network and the regression classification network, completing classification of positive and negative samples, training the anchor-based neural network constructed in the step (1), fixing trained parameters, obtaining two-stage anchor-based neural network models, and performing dynamic video abstraction by using the obtained neural network models. The brand new network structure provided by the invention can process all video frames in parallel, so that the training time is reduced, the final output accuracy is effectively improved, and the network structure is relatively simple and has low requirements on computing power.
Need to check novelty before this filing date? Find Prior Art

Description

technical field

[0001] The present invention relates to a video summarization method, in particular to a two-stage anchor-based dynamic video summarization method. Background technique

[0002] With the development of science and technology, camera equipment has become popular, and the network for uploading and downloading videos has become smoother, followed by an increasing number of video resources. Therefore, this technical requirement has emerged in real life - greatly shortening the length of the video while expressing the main meaning of the original video. The emergence of video summarization methods has successfully responded to this demand. At present, video summarization technology has been successfully applied in many fields, such as surveillance video analysis, generation of movie trailers, motion recognition and so on.

[0003] At present, there are mainly two types of methods for researching video summarization: 1) Static video summarization methods, also kno...

Examples

Embodiment Construction

[0055] Embodiments of the present invention will be disclosed in the following diagrams. For the sake of clarity, many practical details will be described together in the following description. It should be understood, however, that these practical details should not be used to limit the invention. That is, in some embodiments of the invention, these practical details are not necessary.

[0056] The present invention is a two-stage anchor-based dynamic video summarization method, which improves the accuracy and robustness of video summarization through an anchor mechanism and a two-stage network, and the two-stage anchor-based neural network includes a feature The extraction network, the self-attention mechanism network, the two-stage network, namely the regional recommendation network and the regression classification network, specifically include the following steps:

[0057] Step 1: Get the tvsum and summe datasets and preprocess them. tvsum and summe are the most importa...