Unlock instant, AI-driven research and patent intelligence for your innovation.

A method for video recognition and classification based on CNN fusion spatio-temporal salient information

A technology of video recognition and classification method, applied in the field of computer vision, to achieve the effect of improving the accuracy

Active Publication Date: 2019-02-12
BEIJING UNIV OF TECH
View PDF3 Cites 0 Cited by
  • Summary
  • Abstract
  • Description
  • Claims
  • Application Information

AI Technical Summary

Problems solved by technology

This method combines time domain information and spatial domain information, but there is still room for improvement

Method used

the structure of the environmentally friendly knitted fabric provided by the present invention; figure 2 Flow chart of the yarn wrapping machine for environmentally friendly knitted fabrics and storage devices; image 3 Is the parameter map of the yarn covering machine
View more

Image

Smart Image Click on the blue labels to locate them in the text.
Viewing Examples
Smart Image
  • A method for video recognition and classification based on CNN fusion spatio-temporal salient information
  • A method for video recognition and classification based on CNN fusion spatio-temporal salient information
  • A method for video recognition and classification based on CNN fusion spatio-temporal salient information

Examples

Experimental program
Comparison scheme
Effect test

Embodiment Construction

[0017] Such as figure 1 As shown, this video recognition classification method based on CNN fusion spatio-temporal salient information, the method includes the following steps:

[0018] (1) Sampling the video to be identified and classified to obtain a plurality of video clips;

[0019] (2) Each video clip is processed into three sequences: original image sequence, edge image sequence and optical flow image sequence;

[0020] (3) Use the convolutional neural network model to extract features for the three types of data: original image sequence, edge image sequence and optical flow image sequence, and based on these three types of features, calculate the probability that the video clip belongs to different categories;

[0021] (4) fusion of the category probabilities calculated by different features to obtain the classification results of the video clips;

[0022] (5) merging the classification results of each video segment in step (4) to obtain the classification results of ...

the structure of the environmentally friendly knitted fabric provided by the present invention; figure 2 Flow chart of the yarn wrapping machine for environmentally friendly knitted fabrics and storage devices; image 3 Is the parameter map of the yarn covering machine
Login to View More

PUM

No PUM Login to View More

Abstract

The invention discloses a video recognition and classification method based on CNN fusion of spatiotemporal salient information, which can improve the accuracy of video classification. The method comprises steps: (1) sampling video to be identified and classified to obtain a plurality of video clips; (2) processing each video clip into three sequences: original image sequence, edge image sequence and optical flow image sequence; (3) utilizing The convolutional neural network model extracts features for three types of data: original image sequence, edge image sequence, and optical flow image sequence, and based on these three types of features, calculates the probability of video clips belonging to different categories; (4) fused different features to calculate The category probability of the video segment is obtained to obtain the classification result of the video segment; (5) the classification result of each video segment of the fusion step (4) is obtained to obtain the classification result of the video.

Description

technical field [0001] The invention belongs to the technical field of computer vision, and in particular relates to a video recognition and classification method based on CNN fusion of spatiotemporal salient information. Background technique [0002] With the rise of CNN (Covolution Neural Networks, Convolutional Neural Networks) in the field of computer vision, Convolutional Neural Networks have achieved the best results in almost any task related to image classification. In the past two years, the application of convolutional networks in video has gradually increased. The main methods can be divided into three categories: 3D convolutional networks (3-DimensionCovolution Neural Networks, 3DCNN), convolutional networks combined with long-short-term memory (Long-Short TermMemory, LSTM) model and a two-stream method combined with Optical Flow. [0003] The 3D convolutional network method proposed by Ji et al. regards the input video sequence as several three-dimensional bloc...

Claims

the structure of the environmentally friendly knitted fabric provided by the present invention; figure 2 Flow chart of the yarn wrapping machine for environmentally friendly knitted fabrics and storage devices; image 3 Is the parameter map of the yarn covering machine
Login to View More

Application Information

Patent Timeline
no application Login to View More
Patent Type & Authority Patents(China)
IPC IPC(8): G06K9/62G06N3/08
CPCG06N3/08G06F18/24
Inventor 尹宝才王文通王立春孔德慧
Owner BEIJING UNIV OF TECH