Self-acquisition image training data set screening method and device based on deep learning

A training data set and deep learning technology, applied in the field of deep learning and computer vision, can solve the problems of limited recognition speed of feature information recognition models, unsatisfactory data set generation efficiency and accuracy, and inaccurate raw video processing, etc. Network detection, improved screening efficiency and accuracy, and convenient data processing

Pending Publication Date: 2022-07-22
群周科技(上海)有限公司
View PDF2 Cites 0 Cited by
  • Summary
  • Abstract
  • Description
  • Claims
  • Application Information

AI Technical Summary

Problems solved by technology

The disadvantage of this scheme is that the recognition speed of the feature information recognition model is limited due to the large amount of data, and the efficiency and accuracy of data set generation are still not ideal
The disadvantage of this scheme is that the processing of the original video is not accurate enough, resulting in low accuracy of the data set

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
  • Self-acquisition image training data set screening method and device based on deep learning
  • Self-acquisition image training data set screening method and device based on deep learning

Examples

Experimental program
Comparison scheme
Effect test

Embodiment 1

[0049] Take application scenarios where the positions of the camera and a specific target are relatively fixed, such as pedestrian detection, target detection, etc.

[0050] like figure 1 As shown, the self-collected image training dataset screening method based on deep learning includes:

[0051] S100 acquires the original image.

[0052] Specifically, obtaining the original image includes:

[0053] S110 sets the shooting parameters of the camera, and collects original video data through the camera, and the shooting parameters include the frame rate. In general, the shooting parameters also include resolution. Frame rate and resolution are set according to image quality requirements. In this embodiment, the frame rate of the camera is 30 frames / second, and the resolution is 3840*2160.

[0054] The S120 splits the original video data into original images according to the preset frame rate, wherein the preset frame rate is the same as the camera frame rate, that is, the pr...

Embodiment 2

[0071] Take application scenarios where the positions of the camera and a specific target are relatively fixed, such as pedestrian detection, target detection, etc.

[0072] like figure 2 As shown, the self-collected image training data set screening device based on deep learning includes an acquisition module 1 , a cropping module 2 , a detection module 3 and a generation module 4 .

[0073] The acquisition module 1 is used to acquire the original image.

[0074] Specifically, the acquisition module 1 includes an acquisition unit and a deframe unit. The acquisition unit is used to set the shooting parameters of the camera, and collect the original video data through the camera, and the shooting parameters include the frame rate. In general, the shooting parameters also include resolution. Frame rate and resolution are set according to image quality requirements. In this embodiment, the frame rate of the camera is 30 frames / second, and the resolution is 3840*2160. The fr...

Embodiment 3

[0089] An electronic device includes a memory, a processor, and a computer program stored in the memory and running on the processor. Embodiment 1 is implemented when the processor executes the program.

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 self-acquisition image training data set screening method and device based on deep learning, and belongs to the technical field of deep learning and computer vision. In order to solve the problems of low screening efficiency and low precision of a self-acquisition image training data set, the invention provides a self-acquisition image training data set screening method based on deep learning, and the method comprises the steps: obtaining an original image; cutting the original image according to a specific target position to obtain a preprocessed image; detecting and screening the pre-processed image through a plurality of trained neural networks based on deep learning to obtain a pre-processed image set containing the specific target; and generating a training data set according to the set. According to the method, the noise and the calculation amount are reduced through cutting, the screening precision is improved through the trained multiple neural networks based on deep learning, and the screening efficiency and precision of the specific target required in the self-acquired image are greatly improved.

Description

technical field [0001] The invention relates to the technical fields of deep learning and computer vision, and in particular, to a method and device for screening self-collected image training data sets based on deep learning. Background technique [0002] With the continuous development of artificial intelligence technologies such as machine learning and deep learning, artificial intelligence technologies are widely used in various practical scenarios. In the intelligent detection scene for specific targets, the training of vision-based deep learning algorithms requires a large number of self-collected image datasets for model training. Data sets play a key role in improving the accuracy of deep learning algorithms and improving the generalization of models. [0003] At present, the main acquisition method of deep learning data set for self-collected image detection is as follows: collect raw video data through a camera installed on a fixed platform, split the collected im...

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): G06T7/70G06T7/11G06T7/194G06N3/04G06N3/08G06V20/40
CPCG06T7/70G06T7/11G06T7/194G06N3/08G06T2207/10016G06T2207/20084G06T2207/20081G06T2207/20132G06T2207/30244G06N3/045
Inventor 贾宁陈潇曹旺辉
Owner 群周科技(上海)有限公司
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