Human body target detection method and system for real-time application of mobile terminal

A human body target and real-time application technology, applied in the field of artificial intelligence, can solve the problems of human body detection accuracy and poor real-time performance, and achieve the effects of convenient deployment, simple operation, and improved work efficiency and detection accuracy

Pending Publication Date: 2022-03-25
上海震巽智慧科技有限公司
View PDF2 Cites 0 Cited by
  • Summary
  • Abstract
  • Description
  • Claims
  • Application Information

AI Technical Summary

Problems solved by technology

[0006] Aiming at the problems of poor human body detection accuracy and real-time performance at existing mobile terminals, the present invention provides a human body target detection method and system oriented to real-time applications at mobile terminals

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
  • Human body target detection method and system for real-time application of mobile terminal
  • Human body target detection method and system for real-time application of mobile terminal
  • Human body target detection method and system for real-time application of mobile terminal

Examples

Experimental program
Comparison scheme
Effect test

Embodiment 1

[0034] Such as figure 1 As shown, a human target detection method for real-time mobile applications includes the following steps:

[0035] S1: To collect data, the high-speed motion camera can be used as a sensor to capture real-time data and save it as raw data to complete the collection process of different scene data; and preprocess the data. The purpose of preprocessing is to collect the high-speed motion camera Optimize the original data to ensure the quality of the data while reducing the amount of data and improve the speed of data processing; specifically, in the preprocessing step, including bilinear interpolation, anchor point clustering and negative samples for the original data The three processing steps of elimination, bilinear interpolation changes the size of the original data and speeds up the processing speed of the network; anchor point clustering generates corresponding anchor point information suitable for different scenarios, improving network detection ac...

Embodiment 2

[0046] It is basically the same as Embodiment 1. Specifically, in order to further consider the comprehensiveness of the entire process, in this implementation, it focuses on optimizing the network that does not support HiSilicon when deploying HiSilicon as an example. Specifically, in the step S31, when the conversion tool used for format conversion does not support a certain network structure layer of the network model file, the network structure layer can be analyzed in principle, and the network structure can be completed by combining other network structure layers The similar function of the layer realizes the transformation. For example, the activation function Hswish of mobilenetv3 in this application can successfully complete the effect of replacement and transplantation through power+eltwise+relu. When similar functions cannot be completed through combination of other network structure layers, similar functions can be directly implemented in the development board by a...

Embodiment 3

[0048] A human object detection system oriented to real-time applications on mobile terminals, implemented by using the human object detection method oriented to real-time applications on mobile terminals as described in any one of the above-mentioned embodiments 1-2. Specifically, the system includes:

[0049] Acquisition module: used to collect real-time data from high-speed motion cameras;

[0050] Pre-processing module: used to pre-process the real-time data in the acquisition module to ensure data quality and speed up network processing;

[0051] Deep learning network module: used to train the data output by the preprocessing module to obtain the corresponding network model;

[0052] Deployment module: used to deploy the network model output by the deep learning network to the mobile terminal;

[0053] Mobile terminal: used to process the detection results deployed to the mobile terminal and complete the result output.

[0054] The system of the present invention is co...

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 human body target detection method and system for real-time application of a mobile terminal, and belongs to the field of artificial intelligence. Aiming at the problems of poor human body detection precision and poor real-time performance of the existing mobile terminal, the invention provides a human body target detection method for real-time application of the mobile terminal, and the method comprises the following steps: collecting data and carrying out the preprocessing; inputting the preprocessed data into a deep learning framework for training to obtain a network model file; and deploying the network model file to a mobile terminal, completing human body detection and outputting a result. According to the invention, the deep learning framework trains the data to optimize and adapt to the mobile terminal, the subsequent deployment of the mobile terminal is facilitated, the spatial pyramid pooling and the feature pyramid structure can greatly increase the accurate detection of the long-range small target and the shielding condition, and the accurate measurement precision is improved. And when the feature extraction network is deployed to the mobile terminal, the real-time performance can be ensured under the existing computing resources of the equipment, and the whole method is simple to operate and wide in adaptability.

Description

technical field [0001] The invention belongs to the technical field of artificial intelligence, and more specifically relates to a human target detection method and system for real-time mobile applications. Background technique [0002] At present, high-speed motion cameras are more and more widely used in smart driving, smart cities and smart security. In order to ensure the reliability of various application scenarios, it is necessary to accurately detect the human body and increase the detection speed. Due to the complexity of the environment such as cities and roads, human bodies are mutually occluded and the images in the distant view are small, resulting in unsatisfactory detection results of existing detection methods. The advancement of real-time mobile applications such as intelligent driving and intelligent security has constituted a major impact. Existing mobile terminal detection methods are mainly divided into two categories: traditional algorithms and deep lea...

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): G06V40/10G06K9/62G06N3/04G06N3/08G06V10/46G06V10/80
CPCG06N3/08G06N3/045G06F18/253
Inventor 贾宁陈潇刘文强
Owner 上海震巽智慧科技有限公司
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