Target detection training sample screening method

A technology for training samples and screening methods, which is applied in the field of target detection training sample screening to improve efficiency and accuracy.

Active Publication Date: 2021-02-26
ZHEJIANG PECKERAI TECH CO LTD
View PDF3 Cites 0 Cited by
  • Summary
  • Abstract
  • Description
  • Claims
  • Application Information

AI Technical Summary

Problems solved by technology

Each image sample is processed as follows:

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
  • Target detection training sample screening method
  • Target detection training sample screening method
  • Target detection training sample screening method

Examples

Experimental program
Comparison scheme
Effect test

Embodiment 1

[0023] Embodiment 1: In order to solve the above technical problems, this embodiment takes the model of the contraband detection scene as an example to illustrate the sample screening method, including the following steps:

[0024] 1. Use the models of different stages obtained during the training process to detect the targets in the training set data, and obtain the detection results of each image sample on the model M of different stages.

[0025] Among them, the training set is composed of manually labeled image samples. The number of image samples is huge, generally thousands or even hundreds of thousands. The acquisition of image samples is not limited to image acquisition, but also includes samples and / or a new sample obtained through an image fusion method disclosed in the prior art. In addition, there is no limitation on the types of images, which can be obtained by cameras, X-ray security inspection equipment, and terahertz imaging equipment. However, in general, gra...

Embodiment 2

[0037] Example 2: In Example 1 for M 1 -M 13 and M 14 After judging and filtering the forgotten samples to form the set C1, the recognition frame of the 1 -M 5 and M 6 -M 14 The obtained recognition frame is judged and screened to form a set C2 of forgotten samples, and the union of C1 and C2 is eliminated to obtain a new training set. Forgotten samples with different degrees of forgetting can be obtained by screening multiple times in different stages, and the specific number of times can be selected according to the actual effect.

[0038] The technical solution of the present invention can also be applied to target recognition and detection scenarios other than contraband detection in the embodiment, such as face recognition, license plate recognition, road recognition, unmanned driving, lesion detection and analysis in medical image CT inspection scenarios, etc. Various target detection scenarios.

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 discloses a target detection training sample screening method, and the method comprises the steps: detecting a target in training set data through employing models at different stages obtained in a training process, and obtaining a detection result of each image sample on a model M at different stages; and screening the detection results of each image sample in different stage modelsto obtain a complete forgotten sample and a partial forgotten sample. Manual work is replaced by the model to analyze noise samples in a large number of target detection data sets, manpower is saved,the subjective influence of manual data screening is eliminated, and the efficiency and accuracy of executing a target detection task through a deep learning method are improved.

Description

technical field [0001] The invention belongs to the technical field of target detection, and in particular relates to a screening method for target detection training samples. Background technique [0002] In recent years, with the continuous development of artificial intelligence technology, deep learning technology has made breakthroughs in tasks such as classification, recognition, detection, segmentation, and tracking in the field of computer vision. Compared with traditional machine vision methods, deep convolutional neural networks can learn useful features from a large amount of data under the training of large data, which has the advantages of fast speed, high precision, and low cost. However, a large part of the reason why deep learning can achieve this superiority over traditional methods is because deep learning is based on a large amount of data, especially in the field of target detection, deep learning requires a large amount of effective data. In order to pro...

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): G06K9/62G06K9/32G06N3/08
CPCG06N3/08G06V10/25G06V2201/07G06F18/24G06F18/214
Inventor 宋志龙
Owner ZHEJIANG PECKERAI 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