Identification method, model training, system and equipment

A technology for recognition and recognition results, applied in the computer field, can solve problems such as poor algorithm recognition accuracy, and achieve the effect of suppressing noise

Pending Publication Date: 2021-04-13
ALIBABA GRP HLDG LTD
View PDF0 Cites 2 Cited by
  • Summary
  • Abstract
  • Description
  • Claims
  • Application Information

AI Technical Summary

Problems solved by technology

[0004] In the prior art, there are technical solutions for using artificial intelligence algorithms (such as neura

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
  • Identification method, model training, system and equipment
  • Identification method, model training, system and equipment
  • Identification method, model training, system and equipment

Examples

Experimental program
Comparison scheme
Effect test

Embodiment Construction

[0092] In the prior art, there is a video recognition method based on a neural network model, and the existing neural network model usually uses CNN (Convolutional Neural Networks, Convolutional Neural Networks) to extract features corresponding to different frames of the video, and then different All the features corresponding to the frame are input into the fully connected layer, so that the feature fusion is performed by the fully connected layer, and then the final recognition result is obtained. This method directly fuses the features corresponding to different frames through the fully connected layer, and then performs recognition. The effect is poor and affects the recognition accuracy.

[0093] In the process of studying the technical solutions provided by the embodiments of the present application, the inventors found that there are many noises in the features extracted by the feature extraction sub-network, which not only have no effect on the final recognition, but a...

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 embodiment of the invention provides an identification method, model training, system and equipment. The method comprises the following steps: acquiring a to-be-identified video, wherein the to-be-identified video comprises multiple frames of images; inputting a to-be-identified video into the trained neural network model to obtain an identification result, wherein the neural network model is used for performing feature extraction on multiple frames of images to obtain multiple frame features; determining an attention weight corresponding to each element in the plurality of frame features according to the plurality of frame features; multiplying each element in the plurality of frame features by the corresponding attention weight to obtain a plurality of weighted frame features; and integrating the plurality of weighted frame features to obtain the identification result. According to the technical scheme provided by the embodiment of the invention, the noise in the plurality of frame features can be effectively suppressed, the effective features are enhanced, and the accuracy of video recognition can be effectively improved.

Description

technical field [0001] The present application relates to the field of computer technology, in particular to a recognition method, model training, system and equipment. Background technique [0002] Currently, many terminal devices have functions of photographing, filming and / or recording. In some application scenarios, it is necessary to identify the pictures taken by the terminal device, the videos taken and / or the audio data recorded, for example, to classify the pictures, videos or audios. [0003] Taking the smart surveillance camera as an example, the user can set the smart surveillance camera at home to realize the function of watching the nursing home or capturing the wonderful moments of pets. The amount of data of the video or picture captured by the intelligent surveillance camera is very huge, so it is necessary to analyze the video or image captured by the intelligent surveillance camera to filter out data that is not of interest to the user. [0004] In the p...

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
IPC IPC(8): G06K9/00G06N3/04
CPCG06V20/41G06V20/46G06N3/045
Inventor 杨攸奕武元琪李名杨
Owner ALIBABA GRP HLDG LTD
Who we serve
  • R&D Engineer
  • R&D Manager
  • IP Professional
Why Eureka
  • Industry Leading Data Capabilities
  • Powerful AI technology
  • Patent DNA Extraction
Social media
Try Eureka
PatSnap group products

Browse by: Latest US Patents, China's latest patents, Technical Efficacy Thesaurus, Application Domain, Technology Topic.

© 2024 PatSnap. All rights reserved.Legal|Privacy policy|Modern Slavery Act Transparency Statement|Sitemap