Eureka AIR delivers breakthrough ideas for toughest innovation challenges, trusted by R&D personnel around the world.

Image concurrent processing method, device and system based on single GPU card

A technology of a GPU card and a processing method, applied in the field of image processing, can solve the problems of low actual utilization of CPU hardware and waste of resources, and achieve the effects of increasing throughput capacity, avoiding waste of resources, and improving processing efficiency

Inactive Publication Date: 2019-02-26
北京视甄智能科技有限公司
View PDF5 Cites 0 Cited by
  • Summary
  • Abstract
  • Description
  • Claims
  • Application Information

AI Technical Summary

Problems solved by technology

[0005] The purpose of the embodiments of the present invention is to provide a method, device and system for concurrent processing of images based on a single GPU card, to solve the problem that the actual utilization rate of CPU hardware is not high when performing deep learning applications, resulting in waste of resources

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
  • Image concurrent processing method, device and system based on single GPU card
  • Image concurrent processing method, device and system based on single GPU card
  • Image concurrent processing method, device and system based on single GPU card

Examples

Experimental program
Comparison scheme
Effect test

Embodiment 1

[0029] Embodiment 1 of the present invention provides a schematic flowchart of a method for concurrent image processing based on a single GPU card, specifically as figure 1 shown. The method is executed by the task manager, the method includes:

[0030] Step 110, receiving the task submitted by the user.

[0031] Specifically, the tasks submitted by the users are established according to their own processing requirements and objects to be processed. For example, if a user needs to recognize a face image, the submitted tasks may include three tasks: detecting face objects, extracting face features, and judging liveness detection. Processing requirements can actually be understood as task types, and objects to be processed are task data. In addition, users can also define forward dependent tasks when submitting tasks. For example, detecting face objects is a forward-dependent task of extracting facial features, and extracting facial features is a forward-dependent task of ju...

Embodiment 2

[0048] Corresponding to the above-mentioned embodiments, Embodiment 2 of the present invention provides a schematic structural diagram of an image concurrent processing device based on a single GPU card, specifically as figure 2 shown. The device includes: a receiving unit 201 , a configuration unit 202 and a processing unit 203 .

[0049] The receiving unit 201 is configured to receive a task submitted by a user, wherein the task is a task created by the user according to the object to be processed and the processing requirement;

[0050] The configuration unit 202 is configured to add the task to the task queue, and configure a corresponding task thread for the task;

[0051] The processing unit 203 is configured to add the configured task threads to the thread pool in sequence, and the thread pool contains at least two task threads;

[0052] According to the memory of the graphics processor GPU card and the memory occupied by each task thread in the thread pool, reasonab...

Embodiment 3

[0062] Corresponding to the above-mentioned embodiment, the embodiment of the present invention also provides an image concurrent processing system based on a single GPU card, specifically as image 3 As shown, the system includes a task manager 301 and a GPU card 302 .

[0063] The task manager 301 is used to execute the method steps as in Embodiment 1 above, and the GPU card 302 is used to feed back response information corresponding to one or more task threads after receiving one or more task threads submitted by the task manager 301 to the task manager 301; and process one or more task threads submitted by the task manager 301 in parallel. The method steps performed by each component in this embodiment have also been introduced in detail in the above-mentioned embodiment 1, and will not be repeated here.

[0064] An image concurrent processing system based on a single GPU card provided by an embodiment of the present invention, after receiving a task submitted by a user, ...

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 discloses an image concurrent processing method, a device and a system based on a single GPU card. The method includes adding tasks to the task queue and configuring corresponding task threads for the tasks, adding the configured task threads to the thread pool in order; according to the memory of GPU card and the memory occupied by each task thread in the thread pool, selecting one or more task threads reasonably selected from the thread pool and submitted to the GPU card so that the GPU card can process one or more task threads in parallel. By means of the method, the resources of the GPU can be fully utilized, the waste of resources can be avoided, the idle and surplus proportion of the GPU can be greatly reduced, and the throughput of the system can be effectively increased. In addition, multiple task threads can be processed at the same time, which can have a better speed ratio and further improve the processing efficiency.

Description

technical field [0001] The present invention relates to the technical field of image processing, in particular to a method, device and system for concurrent image processing based on a single GPU card. Background technique [0002] Deep learning is a new field in machine learning research. Its motivation is to establish and simulate the neural network of human brain for analysis and learning. It imitates the mechanism of human brain to interpret data, such as images, sounds and texts. [0003] Currently, the most computationally important method for deep learning applications is forward propagation. Under the demands of a large number of application scenarios today, the requirements for deep learning computing performance are getting higher and higher. The most commonly used method for improving computing performance is to use a graphics processing unit (Graphics Processing Unit, referred to as GPU) to accelerate computing, so as to achieve higher system throughput. The cl...

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): G06F9/50G06F9/48
CPCG06F9/4881G06F9/5027G06F2209/5011G06F2209/5018
Inventor 安玉山
Owner 北京视甄智能科技有限公司
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
Eureka Blog
Learn More
PatSnap group products

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

© 2024 PatSnap. All rights reserved.Legal|Privacy policy|Modern Slavery Act Transparency Statement|Sitemap|About US| Contact US: help@patsnap.com