Training method, image quality detection method, device and medium

A training method and quality technology, applied in the field of computer vision, can solve the problems of reducing the performance of quality detection models, large labor costs and time costs, low label accuracy, etc., to save labeling costs, improve accuracy, and avoid higher accuracy. low effect

Active Publication Date: 2022-08-05
SHENZHEN MICROBT ELECTRONICS TECH CO LTD
View PDF12 Cites 0 Cited by
  • Summary
  • Abstract
  • Description
  • Claims
  • Application Information

AI Technical Summary

Problems solved by technology

[0004] In practical applications, the annotation of labeled image samples usually consumes a lot of human and time costs
Moreover, for the same image sample, different people often give different labels, which leads to low accuracy of the label of the image sample; and the performance of the quality detection model is closely related to the accuracy of the label, so the low accuracy of the label Also reduces the performance of the quality detection model

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
  • Training method, image quality detection method, device and medium
  • Training method, image quality detection method, device and medium
  • Training method, image quality detection method, device and medium

Examples

Experimental program
Comparison scheme
Effect test

Embodiment 1

[0075] refer to figure 2 , showing a schematic flowchart of the steps of a training method according to an embodiment of the present application, and the method may specifically include the following steps:

[0076] Step 201, determine the standard image containing the target;

[0077] Step 202, determining the test image corresponding to the standard image;

[0078] Step 203, using the target recognition model to determine the degree of matching between the standard image and the test image;

[0079] Step 204, according to the matching degree, determine the label of the test image;

[0080] Step 205: Train the quality detection model according to the test image and the label of the test image; the quality detection model is used to determine the quality score corresponding to the test image; in the training process of the quality detection model, according to the label and the quality score; The quality score determines loss information, and the parameters of the quality ...

Embodiment 2

[0153] In order to improve the convergence speed of the quality detection model (loss information meets the first preset condition), the embodiment of the present application may first use the target recognition task to pre-train the feature extraction unit of the quality detection model; Using the quality detection task, the feature extraction unit and the quality score determination unit of the quality detection model are trained until the quality detection task converges (the loss information meets the first preset condition).

[0154] refer to Figure 5 , showing a schematic flowchart of a training method according to an embodiment of the present application, wherein, in the pre-training stage, a target identification unit may be connected after the feature extraction unit. A feature extraction unit and an object recognition unit may be used to perform object recognition tasks. The training data for the target recognition task may include: a training image set for target ...

Embodiment 3

[0168] This embodiment describes the quality detection process of the quality detection model. The quality detection model may perform quality detection on the image to be recognized to obtain a corresponding quality score.

[0169] refer to Figure 7 , showing a schematic flowchart of the steps of a quality detection method according to an embodiment of the present application, and the method may specifically include the following steps:

[0170] Step 701, receiving an image to be recognized;

[0171] Step 702, using a quality detection model to determine the quality score corresponding to the to-be-recognized image;

[0172] Wherein, the training process of the quality detection model may specifically include: determining a standard image containing a target; determining a test image corresponding to the standard image; using a target recognition model to determine the degree of matching between the standard image and the test image; Matching degree, determine the label of...

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 provides a training method, an image quality detection method and device and a medium. The training method specifically comprises the following steps: determining a standard image containing a target; determining a test image corresponding to the standard image; determining a matching degree between the standard image and the test image by using a target recognition model; determining a label of the test image according to the matching degree; training a quality detection model according to the test image and the label of the test image; the quality detection model is used for determining a quality score corresponding to the test image; in the training process of the quality detection model, loss information is determined according to the label and the quality score, and parameters of the quality detection model are updated according to the loss information. According to the embodiment of the invention, the marking cost of the test image can be saved, and the accuracy and other performances of the quality detection model can be improved.

Description

technical field [0001] The present application relates to the technical field of computer vision, and in particular, to a training method, an image quality detection method, an apparatus and a medium. Background technique [0002] In the field of computer vision technology, target recognition models can be used to recognize targets such as people, animals, vehicles, and characters contained in images. In practical applications, the target is far away from the camera, the lighting conditions are poor, and the imaging effect of the image acquisition device is poor, which may easily lead to poor image clarity and low resolution, thus affecting the target recognition effect. In order to improve the target recognition results, the image quality can be checked before target recognition, and the images with poor quality can be blocked from entering the target recognition model. [0003] The current image quality detection methods usually firstly label the image samples manually to...

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): G06V10/774G06V10/74G06V10/82G06N20/00
CPCG06V10/761G06V10/774G06V10/82G06N20/00
Inventor 艾国凌明杨作兴
Owner SHENZHEN MICROBT ELECTRONICS TECH 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