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

Intelligent quality inspection method and system for image annotation data

An image labeling and data technology, applied in still image data retrieval, still image data clustering/classification, neural learning methods, etc. The effect of performance indicators

Pending Publication Date: 2022-03-15
WUHAN ZHONGHAITING DATA TECH CO LTD
View PDF0 Cites 1 Cited by
  • Summary
  • Abstract
  • Description
  • Claims
  • Application Information

AI Technical Summary

Problems solved by technology

[0004] In order to solve the problem of high cost and low efficiency of manual labeling data set quality inspection, in the first aspect of the present invention, an intelligent quality inspection method for image labeling data is provided, including: obtaining multiple labeling images and corresponding labeling files; Use the pre-training model to predict each labeled image to obtain the predicted results of one or more targets; compare the predicted results of the one or more targets with the marked files of each marked image, Consistency between the comparison and category judgments; correcting the labeling files judged to be inconsistent, and using the corrected labeling files to train the pre-training 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
  • Intelligent quality inspection method and system for image annotation data
  • Intelligent quality inspection method and system for image annotation data
  • Intelligent quality inspection method and system for image annotation data

Examples

Experimental program
Comparison scheme
Effect test

Embodiment 2

[0039] refer to Figure 5 , the second aspect of the present invention provides an intelligent quality inspection system 1 for image annotation data, including: an acquisition module 11 for obtaining multiple annotation images and their corresponding annotation files; a prediction module 12 for using the predicted The training model predicts each marked image to obtain the predicted result of one or more targets; the judging module 13 is used to compare the predicted result of the one or more targets with the marked file of each marked image, according to each The consistency between the intersection and union ratio of each target and the category judgment; the training module 14 is used to correct the label files that are judged to be inconsistent, and use the corrected label files to train the pre-training model.

[0040] Further, the judging module 13 includes an acquisition unit, an analysis unit and a judgment unit, the acquisition unit is used to acquire the annotation f...

Embodiment 3

[0042] refer to Figure 6 , the third aspect of the present invention provides an electronic device, including: one or more processors; storage means for storing one or more programs, when the one or more programs are used by the one or more executed by one or more processors, so that the one or more processors implement the method of the first aspect of the present invention.

[0043] The electronic device 500 may include a processing device (such as a central processing unit, a graphics processing unit, etc.) 501, which may be loaded into a random access memory (RAM) 503 according to a program stored in a read-only memory (ROM) 502 or loaded from a storage device 508 Various appropriate actions and processing are performed by the program. In the RAM 503, various programs and data necessary for the operation of the electronic device 500 are also stored. The processing device 501 , ROM 502 and RAM 503 are connected to each other through a bus 504 . An input / output (I / O) int...

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 relates to an intelligent quality inspection method and system for image annotation data. The method comprises the following steps: acquiring a plurality of annotation images and annotation files corresponding to the annotation images; predicting each labeled image by using a pre-training model to obtain a prediction result of one or more targets; comparing the prediction result of the one or more targets with the annotation file of each annotation image, and judging the consistency of the prediction result and the annotation file according to the intersection-union ratio and the category of each target; and correcting the marked files which are judged to be inconsistent, and training the pre-training model by utilizing the corrected marked files. According to the method, quality inspection is carried out on the image annotation file through the pre-training model, IOU calculation and category information, errors of the annotation file can be rapidly and accurately detected, and meanwhile the generalization ability and defects of the model to a training data set can be evaluated; and finally, retraining the network model by using the updated annotation data, thereby improving various performance indexes of the model.

Description

technical field [0001] The invention belongs to the technical field of deep learning and image processing, and in particular relates to an intelligent quality inspection method and system for image labeling data. Background technique [0002] The model obtained by deep learning training is obtained by continuous iteration of the deep learning algorithm and the training data set. When the developer selects a suitable algorithm, the index of the deep learning model depends to a large extent on the selection of parameters and labeling data of. [0003] However, the annotation data set of the most original training model is manually annotated, so the obtained annotation results will inevitably have some errors. When the amount of annotated data is large or there are many types of annotations, manual quality inspection and annotation of data will be time-consuming. It is a labor-intensive process, but the accuracy of standard labeling data obtained through quality inspection is ...

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): G06F16/55G06N3/04G06N3/08
CPCG06F16/55G06N3/08G06N3/045
Inventor 何云何豪杰喻旸
Owner WUHAN ZHONGHAITING DATA TECH CO LTD