Neural network chip realization method used for real-time image identification

A neural network and neural network technology, applied in the field of image recognition chips, can solve problems such as large resource requirements and poor real-time performance, and achieve the effects of good scalability, reduced development difficulty, and improved development efficiency

Inactive Publication Date: 2016-09-21
INSPUR GROUP CO LTD
View PDF3 Cites 18 Cited by
  • Summary
  • Abstract
  • Description
  • Claims
  • Application Information

AI Technical Summary

Problems solved by technology

[0005] Aiming at the problems of large resource requirements and poor real-time performance of the real-time image processing system in the prior art, the present invention provides a neural network chip implementation method for real-time image recognition, which can greatly improve the neural network chip for real-time image recognition development efficiency

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
  • Neural network chip realization method used for real-time image identification
  • Neural network chip realization method used for real-time image identification

Examples

Experimental program
Comparison scheme
Effect test

Embodiment Construction

[0025] A neural network chip implementation method for real-time image recognition:

[0026] Set up an automatic compilation platform, the convolutional neural network for image recognition forms a neural network model library,

[0027] According to the application requirements, use the automatic compilation platform to complete the training of the neural network, extract the structure and information of the neural network, and use the information to form and reconstruct the control data of the FPGA.

[0028] Construction of FPGA implementation: According to the received control data of the compiling platform, complete the connection and quick configuration of each computing unit inside the FPGA, complete the connection of the control signal and data signal path, and quickly input the computing data into each computing unit,

[0029] Generation of neural network chip: According to the neural network that needs to be configured, use the image that needs to be recognized for tra...

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 present invention discloses a neural network chip realization method used for real-time image identification, and belongs to the image identification chip technology field. The method comprises the steps of selecting and configuring a neural network model conveniently and rapidly according to the actual demands; realizing a universal neural network chip based on an FPGA, and finishing the neural network generation in the chip according to the FPGA reconstruction data provided by a compiling platform at the later period, so that a hardware platform can realize various neural networks conveniently and rapidly. The method enables the development efficiency of the neural network chip used for the real-time image identification to be improved substantially, the chip development difficulty to be reduced greatly and the project promotion speed to be improved, and enables the chip to have the characteristics of being good in expansibility, being flexible to modify, etc.

Description

technical field [0001] The invention discloses a method for realizing a neuron network chip, which belongs to the technical field of image recognition chips, in particular to a method for realizing a neuron network chip for real-time image recognition. Background technique [0002] At present, due to the need for real-time processing of on-site images, machine vision technology is increasingly implemented with the help of hardware, such as DSP chips, dedicated image signal processing cards, etc. However, DSP image processing also faces the problem of slow processing speed and poor real-time performance of the system due to the large amount of data storage and processing, and the pattern of the original image recognition algorithm cannot recognize some changes in the picture. Target, poor versatility, low recognition success, limited scope of use, unable to complete functions such as image translation, object recognition, and blurred face recognition. [0003] The invention ...

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): G06N3/06G06K9/00
CPCG06V10/94G06N3/065
Inventor 赵鑫鑫姜凯李朋
Owner INSPUR GROUP CO 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