Image feature extraction method and device and electronic equipment

An image feature and extraction method technology, applied in the field of image processing, can solve problems such as slow speed and loss of accuracy, and achieve the effects of avoiding loss of accuracy, improving accuracy and improving computing performance

Pending Publication Date: 2022-01-07
MEGVII BEIJINGTECH CO LTD +1
View PDF0 Cites 1 Cited by
  • Summary
  • Abstract
  • Description
  • Claims
  • Application Information

AI Technical Summary

Problems solved by technology

The inventor found through research that when most existing processors use this method for feature extraction, not only will the accuracy of feature extraction be lost, but also there is still a problem of slow speed

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 feature extraction method and device and electronic equipment
  • Image feature extraction method and device and electronic equipment
  • Image feature extraction method and device and electronic equipment

Examples

Experimental program
Comparison scheme
Effect test

Embodiment 1

[0027] First, refer to figure 1 An example electronic device 100 for implementing a method and apparatus for extracting image features according to an embodiment of the present invention will be described.

[0028] Such as figure 1 Shown is a schematic structural diagram of an electronic device. The electronic device 100 includes one or more processors 102, one or more storage devices 104, an input device 106, an output device 108, and an image acquisition device 110. These components pass through a bus system 112 and / or other forms of connection mechanisms (not shown). It should be noted that figure 1 The components and structures of the electronic device 100 shown are only exemplary, not limiting, and the electronic device may have figure 1 Some components shown may also havefigure 1 Other components and structures are not shown.

[0029] The processor 102 can be implemented in at least one hardware form of a digital signal processor (DSP), a field programmable gate arra...

Embodiment 2

[0036] see figure 2 A schematic flow chart of a method for extracting image features is shown, the method is applied to the above-mentioned electronic device, specifically, it can be executed by a processor in the electronic device, and the method mainly includes the following steps S202 to S204:

[0037] Step S202, acquiring an image to be processed and a preset convolution kernel.

[0038] Wherein, the image to be processed includes an original image or a feature map, and the data types of the image to be processed and the convolution kernel are fixed-point types. The original image can include the initial image captured by the image acquisition device, downloaded from the network, locally stored or manually uploaded, such as an RGB image, and the feature map can include the next layer obtained after convolution operation on the initial image or the intermediate feature map feature map. Wherein, the fixed-point type may include int8 type, int16 type, or int32 type, etc., ...

Embodiment 3

[0065] On the basis of the foregoing embodiments, this embodiment provides a specific example of applying the foregoing image feature extraction method, see Figure 4 A specific schematic diagram of convolution processing shown in the embodiment of the present invention uses floating-point calculations for input conversion, weight conversion, batch matrix multiplication and output conversion, and the data types of the input conversion matrix, weight conversion matrix and output conversion matrix are all Take the floating-point type as an example to illustrate.

[0066]In a specific implementation, the data types of the image to be processed and the convolution kernel are both fixed-point and int8 types, and the input conversion matrix of the floating-point type is used to perform input conversion on the image to be processed (the calculation method adopted is floating point calculation) to obtain the input conversion result of the floating point type, and use the weight conver...

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 an image feature extraction method and device and electronic equipment, and relates to the technical field of image processing, and the method comprises the steps of obtaining a to-be-processed image and a preset convolution kernel, wherein the to-be-processed image comprises an original image or a feature map, and the data types of the to-be-processed image and the convolution kernel are both fixed-point types; performing convolution processing on the to-be-processed image based on the convolution kernel to obtain a convolution processing result used for representing image features of the to-be-processed image, wherein the data type of the convolution processing result is a fixed point type; involving floating point calculation and conversion between fixed points and floating points in the convolution processing process. The invention can effectively improve the feature extraction precision and speed.

Description

technical field [0001] The present invention relates to the technical field of image processing, in particular to an image feature extraction method, device and electronic equipment. Background technique [0002] In the field of computer vision (CV for short), deep learning methods are widely used. Among them, since convolution is an important operator in deep learning methods, it is mainly used for image feature extraction. The computational complexity of the deep learning method is largely affected by the performance of the convolution calculation. In order to speed up the calculation, quantization is usually used in the convolution process, that is, the data involved in the convolution process is They are all quantized to fixed points, and then the plastic calculation method is used in the data processing process. The inventors have found through research that when most existing processors perform feature extraction in this way, not only the accuracy of feature extractio...

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): G06V10/46G06F17/15G06F17/16G06F7/57
CPCG06F17/15G06F17/16G06F7/57
Inventor 陈其友吴博曾平许欣然
Owner MEGVII BEIJINGTECH 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