Method for generating video abstract on basis of deep learning technology

Abstract

The invention discloses a method for generating video abstract on the basis of a deep learning technology. The method includes modeling backgrounds of video stream frame by frame and acquiring moving foregrounds to be used as candidate moving objects; tracking the candidate moving objects of each frame by the aid of a multi-object tracking algorithm and updating candidate objects which form movement tracks; training object classifiers by the aid of convolutional neural networks, confirming the candidate objects and determining categories of the objects by the aid of the classifiers after real moving objects are confirmed; fitting all the real moving objects and relevant information on a small quantity of images, forming video snapshots and displaying the video snapshots to users. The method has the advantages that the real objects and noise can be accurately differentiated from one another by the aid of the deep learning technology; the objects do not need to be confirmed frame by frame owing to an accurate multi-object tracking technology, accordingly, the computational complexity can be greatly reduced, an omission factor of the objects and a false alarm rate of the noise can be effectively reduced, the video processing speeds can be increased, and the method can be applied to various complicated scenes.

Description

technical field [0001] The present invention relates to the technical field of image processing, and more specifically, to a video summary generation method based on deep learning technology. Background technique [0002] In modern society, video surveillance systems play an important role in all walks of life, and play an important role in maintaining social order, strengthening social management and security; however, with the rapid increase in the number of cameras, the storage of massive surveillance video data and the understanding of events recorded in these videos will consume a lot of manpower and material resources. According to the statistics of ReportLinker, in 2011, there were more than 165 million surveillance cameras in the world, generating 1.4 trillion hours of surveillance data. If 20% of the important surveillance video data needs to be watched manually, more than 100 million laborers will be employed (8 hours a day, 300 days a year). Therefore, condensing ...

AI Technical Summary

Problems solved by technology

[0004] For surveillance video, the shooting scene of surveillance video is very complicated: in some scenes, there are many vehicles and the movement speed is fast, such as a highway; in some scenes, the moving target occupies a small pixel area on the screen; in some scenes, Uninterested objects such as trees and flags also move due to the wind, etc.; the complexity of the scene brings great challenges to the accurate detection of moving targets

The current video summarization technology cannot solve the detection problem of moving objects in complex scenes very well, and often makes the missing detection rate of moving objects very high. video summaries for omit important information from the original video

Images