MobileNetV3-based real-time human head classification detection method

A technology of classification detection and human head, which is applied in the field of face recognition, can solve the problems of face recognition tracking error, high cost of human head detection, low precision, etc., and achieve the effect of simple structure, reliable design principle and wide application prospect

Active Publication Date: 2020-10-27
星宏集群有限公司
View PDF6 Cites 11 Cited by
  • Summary
  • Abstract
  • Description
  • Claims
  • Application Information

AI Technical Summary

Problems solved by technology

[0013] Aiming at the above-mentioned defects of traditional face recognition tracking due to the influence of wearing a mask in the prior art, and directly adding head detection to the existing face recognition, the defects of high cost, low precision and no classification, the present invention provides a real-time head detection based on MobileNetV3. Classification detection method to solve the above technical problems

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

Examples

Experimental program
Comparison scheme
Effect test

Embodiment 1

[0051] Such as figure 1 As shown, the present invention provides a kind of real-time head classification detection method based on MobileNetV3, comprises the steps:

[0052] S1. Select a training data set and classify and mark the selected training data set; the training data set includes a mask face data set, a human face data set and a head data set;

[0053] S2. Perform enhanced processing on the image data of the labeled training data set;

[0054] S3. Generate a priori box for each category according to different categories of the training data set;

[0055] S4. Build a model based on MobileNetV3, and then train the model through the training data set;

[0056] S5. Apply the trained model on the large screen, reason the real-time head image captured by the large screen camera, and obtain the classification and detection results.

Embodiment 2

[0058] Such as figure 2 As shown, the present invention provides a kind of real-time head classification detection method based on MobileNetV3, comprises the steps:

[0059] S1. Select a training data set and classify and mark the selected training data set; the training data set includes a mask face data set, a face data set and a head data set; the specific steps are as follows:

[0060] S11. select the training data set; select the mask face data set of the MAFA type, the face data set of the Widerface type and the head data set of the SCUT-HEADv1.0 type;

[0061] S12. Mark the selected training data set, mark the mask face data set of MAFA type as maskface, mark the face data set of Widerface type as face, and mark the head data set of SCUT-HEADv1.0 type as head ; The original MAFA only marked part of the face of the mask, the original Widerface only marked the face, although the original SCUT-HEADv1.0 marked all the heads that appeared, but the labels are the same, with...

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 provides a MobileNetV3-based real-time human head classification detection method. The method comprises the following steps of S1, selecting a training data set and performing classification labeling on the selected training data set, wherein the training data set comprises a mask face data set, a face data set and a head data set; S2, performing enhancement processing on the image data of the labeled training data set; S3, generating a priori box for each category according to different categories of the training data set; S4, constructing a model based on the MobileNetV3, and training the model through the training data set; and S5, applying the trained model to the large screen, and reasoning the real-time human head image captured by the large screen camera to obtain a classification detection result. According to the invention, rapid and accurate classification detection of the human head is realized, key information points of the human face are provided, and a classification detection result provides a data basis for subsequent processing.

Description

technical field [0001] The invention belongs to the technical field of face recognition, and in particular relates to a real-time head classification and detection method based on MobileNetV3. Background technique [0002] MobileNetV3, a small and efficient network structure proposed by Google for mobile or embedded device applications. [0003] FPN, short for Feature Pyramid Net, feature pyramid network. [0004] SSH, the abbreviation of Single Stage Headless Face Detector, single-step face recognition. [0005] At present, the accuracy and speed of face recognition technology can meet the requirements of the application, and the tracking and statistics products based on face recognition have been implemented. However, due to the impact of the epidemic, the following problems have emerged: [0006] (1) Wearing a mask makes the tracking statistics based on face recognition wrong; [0007] (2) The statistics based on head detection generally only count the flow of people, ...

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 Applications(China)
IPC IPC(8): G06K9/00G06K9/62G06N3/04G06N3/08
CPCG06N3/08G06V40/172G06N3/045G06F18/214G06F18/2414G06F18/25
Inventor 冯希宁
Owner 星宏集群有限公司
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