Looking for breakthrough ideas for innovation challenges? Try Patsnap Eureka!

Image processing method and device and target detection model training method and system

An image processing and image technology, applied in the field of image processing and model training, can solve the problems of low model training accuracy, unreasonable processing methods, high difficulty, etc., to avoid low accuracy or excessive difficulty, improve robustness, Guaranteed balanced effect

Pending Publication Date: 2021-07-13
BEIJING WENAN INTELLIGENT TECH CO LTD
View PDF0 Cites 1 Cited by
  • Summary
  • Abstract
  • Description
  • Claims
  • Application Information

AI Technical Summary

Problems solved by technology

[0007] The main purpose of the present invention is to provide an image processing method and device as well as a target detection model training method and system to solve the problem of unreasonable processing methods for original images with large pixel sizes in the prior art, which ultimately leads to the accuracy of model training. Problems that are too low or too difficult

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 processing method and device and target detection model training method and system
  • Image processing method and device and target detection model training method and system
  • Image processing method and device and target detection model training method and system

Examples

Experimental program
Comparison scheme
Effect test

Embodiment 1

[0063] For the target object-pedestrian, in this embodiment, the pixel height corresponding to the pedestrian's clipping threshold range is [40,160], and the first preset scaling ratio r is calculated by the formula (2): r=P / Q, where the pedestrian belongs The pixel height corresponding to the category takes a value of 119 in the clipping threshold range, and the actual measured pixel height Q of pedestrians in the first original image is 131, then the calculated first preset scaling ratio r is 0.9084; therefore, for figure 2 The first original image of is scaled according to the first preset scaling ratio r equal to 0.9084 to obtain image 3 The first image to be cut, such as image 3 As shown, the part selected by the white wire frame is the cropped positive sample image block, which contains a complete pedestrian. In this embodiment, the pixel size of the positive sample image block is 512×512, and its cropping position is random. exist figure 2 Among them, the largest ...

Embodiment 2

[0065] For the target object-non-motor vehicle, in this embodiment, the pixel height corresponding to the non-motor vehicle clipping threshold range is [60,180], and the first preset scaling ratio r is calculated by formula (2): r=P / Q , take a value of 136 in the pixel height clipping threshold range corresponding to the category of non-motor vehicles, figure 2 The actual measured pixel height Q of the non-motor vehicle (the bicycle in the lower left corner of the figure) in the first original image is 337, and the first preset scaling ratio r is calculated to be 0.4036; therefore, for figure 2 The first original image of is scaled according to the first preset scaling ratio r equal to 0.4036 to obtain Figure 5 The first image to be cut, such as Figure 5 As shown, the part selected by the white wire frame is the cropped positive sample image block, which contains a complete non-motorized vehicle (bicycle). In this embodiment, the pixel size of the positive sample image bl...

Embodiment 3

[0067] For the target object-non-motor vehicle, in this embodiment, the pixel height corresponding to the non-motor vehicle clipping threshold range is [60,180], and the first preset scaling ratio r is calculated by formula (2): r=P / Q , take a value of 82 in the pixel height clipping threshold range corresponding to the category of non-motor vehicles, figure 2 The actual measured pixel height Q of the non-motor vehicle in the middle of the picture (the electric bicycle on the upper middle of the figure) in the first original image is 160, then the calculated first preset scaling ratio r is 0.5125; therefore, for figure 2 The first original image of is scaled according to the first preset scaling ratio r equal to 0.5125 to obtain Figure 7 The first image to be cut, such as Figure 7 As shown, the part selected by the white wire frame is the cropped positive sample image block, which contains a complete non-motorized vehicle (electric bicycle). In this embodiment, the pixel ...

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 processing method and device and a target detection model training method and system. The image processing method comprises the steps of counting the number of all types of target objects displayed in a first original image, and calculating the corresponding resampling frequency according to the number of all types of target objects; obtaining a first preset scaling ratio, scaling the first original image according to the first preset scaling ratio to obtain a first to-be-cut image, and for each target object in each category of target objects, cutting a positive sample image block completely containing the target object from the first to-be-cut image; and collecting all cut positive sample image blocks into a sample image pool to serve as a first type of to-be-trained sample images. According to the invention, the problem that the precision of model training is too low or the difficulty is too high due to the unreasonable processing method of the original image with the large pixel size in the prior art is solved.

Description

technical field [0001] The present invention relates to the technical field of image processing and model training, in particular to an image processing method and device, and a target detection model training method and system. Background technique [0002] With the continuous development of imaging technology, the clarity of video or images captured by imaging equipment is gradually improved, but the problem that comes with it is that the pixel size of video or images is getting larger and larger, so that the GPU processing it Memory requirements are getting higher and higher. [0003] When training a deep learning model, considering model training efficiency and economical cost control, it is usually necessary to process an image with a large pixel size first, and then input the processed image as a sample image into the deep learning model for training . [0004] In related technologies, the original image processing method with larger pixel size: [0005] One is to d...

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): G06K9/62G06N20/00
CPCG06N20/00G06V2201/07G06F18/214
Inventor 陈映曹松任必为宋君陶海
Owner BEIJING WENAN INTELLIGENT TECH CO LTD
Who we serve
  • R&D Engineer
  • R&D Manager
  • IP Professional
Why Patsnap Eureka
  • Industry Leading Data Capabilities
  • Powerful AI technology
  • Patent DNA Extraction
Social media
Patsnap 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