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

Efficient multi-channel video analysis method based on heterogeneous parallel acceleration

A technology of multi-channel video and analysis method, which is applied in the field of image and video processing, and can solve the problems of waste of GPU resources and short processing time.

Inactive Publication Date: 2020-06-12
成都鼎安华智慧物联网股份有限公司
View PDF6 Cites 1 Cited by
  • Summary
  • Abstract
  • Description
  • Claims
  • Application Information

AI Technical Summary

Problems solved by technology

However, many of the current video processing methods in practical applications are based on the architecture of CPU serial operation, and can only process one channel of video serially.
[0004] Although there are already some algorithms that combine GPU and CPU and use deep learning to improve the accuracy of video image processing, they all have some significant disadvantages: the processing time of the GPU algorithm is much shorter than that of the CPU in the preparation stage. , the waiting time of the GPU will inevitably cause a waste of GPU 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
  • Efficient multi-channel video analysis method based on heterogeneous parallel acceleration
  • Efficient multi-channel video analysis method based on heterogeneous parallel acceleration
  • Efficient multi-channel video analysis method based on heterogeneous parallel acceleration

Examples

Experimental program
Comparison scheme
Effect test

Embodiment Construction

[0021] The present invention will be further described below in conjunction with accompanying drawing.

[0022] An efficient multi-channel video analysis method based on heterogeneous parallel acceleration, including the following modules and steps:

[0023] Such as figure 1 Among them, the present invention includes following three modules:

[0024] Step 1) main program: mainly responsible for reading and storing each video stream data into the specified shared memory queue;

[0025] Step 2) Algorithm program: mainly responsible for reading video frames from the shared memory queue and assembling them into batch data for accelerated algorithm processing, and finally putting the processing results into the shared memory queue B;

[0026] Step 3) Structure post-processing program: responsible for reading the result from the shared memory B and distributing it to the specified interface.

[0027] The main program described in the above-mentioned steps 1) refers to: read the v...

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 an efficient multi-channel video analysis method based on heterogeneous parallel acceleration, which uses heterogeneous parallel acceleration equipment to accelerate video processing and improve the utilization rate of GPU parallel equipment at the same time. By utilizing a mechanism that multiple processes share a memory queue, waiting time generated by speed difference between parallel equipment and a CPU (Central Processing Unit) is reduced, so that the processing efficiency and the utilization rate of parallel acceleration equipment such as a GPU (Graphics Processing Unit) are improved.

Description

technical field [0001] The invention relates to a multi-channel video analysis method, in particular to a multi-channel video analysis framework method supported by heterogeneous parallel hardware, belonging to the field of image and video processing. Background technique [0002] With the rise and application of parallel devices such as GPU and FPGA, the design method of task parallelism becomes more and more popular. Especially in recent years, algorithms such as machine learning and deep learning have flourished with the support of parallel hardware. Image and video processing tasks are typical tasks in machine learning and deep learning in recent years and have achieved very remarkable results. [0003] Many image processing methods based on deep learning can process image data of a batch containing multiple images at one time. However, many current video processing methods in practical applications are based on the CPU serially running architecture, and can only proce...

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): G06F9/54
CPCG06F9/544G06F9/546
Inventor 庄永忠廖长明徐燕生
Owner 成都鼎安华智慧物联网股份有限公司