Object detection model training method and device, object detection model detection method and device, equipment and medium

An object detection and model training technology, applied in the field of image processing, can solve the problems of high detection rate, complex network design, low detection rate, etc., and achieve the effect of improving the detection rate

Pending Publication Date: 2020-05-12
REMO TECH CO LTD
View PDF14 Cites 5 Cited by
  • Summary
  • Abstract
  • Description
  • Claims
  • Application Information

AI Technical Summary

Problems solved by technology

In the application process, there are often situations where the detection rate of objects of a certain scale is high and the detection rate of objects of other scales is low, and it cannot be guaranteed that objects of any scale can be effectively detected.
In order to solve this problem, existing target detection methods consider introducing the same semantic features for models of different scale objects or designing ...

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
  • Object detection model training method and device, object detection model detection method and device, equipment and medium
  • Object detection model training method and device, object detection model detection method and device, equipment and medium
  • Object detection model training method and device, object detection model detection method and device, equipment and medium

Examples

Experimental program
Comparison scheme
Effect test

Embodiment 1

[0035] Please refer to figure 1 , this embodiment provides a method for training an object detection model, as shown in the figure, the method includes the following steps:

[0036] Step S101. Obtain an initial network model, where the initial network model includes at least two detection sub-modules for objects in different scale intervals.

[0037] According to the predefined network structure and training method, an initial network model is obtained. The initial network model is a conventional object detection model. For example, the SSD algorithm based on the VGG network structure is obtained according to the training method of the SSD algorithm. A network model, in which different detection sub-modules are created for different scale intervals in the initial network model. For the division of the scale interval, as an example, the scale can be divided into three scale intervals [0.00001,0.3), [0.3,0.6) and [0.6,1.0], then the corresponding initial network model also incl...

Embodiment 2

[0044] Please refer to figure 2 , this embodiment provides a method for training an object detection model, as shown in the figure, the method includes the following steps:

[0045] Step S201. Obtain an initial network model, where the initial network model includes at least two detection sub-modules for objects in different scale intervals.

[0046] The scale is divided into at least two scale intervals, and a detection sub-module corresponding to each scale interval is created in the initial network model obtained through conventional methods.

[0047] Step S202a, select a scale interval.

[0048] Step S202b. Randomly select a training sample image, and calculate the area ratio of the marked objects in the training sample image.

[0049] Randomly select a training sample image, obtain the labeled object in the training sample image, and calculate the area ratio of the labeled object in the training sample image. In this embodiment, the object is marked in the form of a bou...

Embodiment 3

[0068] Please refer to image 3 , this embodiment provides a method for training an object detection model, as shown in the figure, the method includes the following steps:

[0069] Step S301. Obtain an initial network model, where the initial network model includes at least two detection sub-modules for objects in different scale intervals.

[0070] Step S302, adjusting the corresponding data augmentation strategy according to the scale interval.

[0071] For the content in this step, please refer to step S102 in the first embodiment and steps S202a-S202g in the second embodiment to obtain a more accurate data augmentation strategy corresponding to each scale interval after adjustment, which will not be repeated here.

[0072] Step S303a, select a scale interval.

[0073] Step S303b, using the adjusted data augmentation strategy corresponding to the scale interval to augment the training sample image of the marked object, so that the size of the marked object is within the ...

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 an object detection model training method and device, an object detection model detection method and device, equipment and a medium. The object detection model training methodcomprises the steps of obtaining an initial network model, wherein the initial network model comprises at least two detection sub-modules for objects in different scale intervals; adjusting a corresponding data augmentation strategy according to the scale interval; and augmenting the training sample images by using the adjusted data augmentation strategy, and training the corresponding detection sub-modules in the initial network model by using the augmented training sample images according to the scale interval to obtain an object detection model. According to the method, the network complexity does not need to be increased, the data diversity is increased by adjusting the data augmentation strategy, and the detection rate of the object detection model obtained by training can be effectively improved.

Description

technical field [0001] The present invention relates to the technical field of image processing, in particular to an object detection model training method, detection method, device, equipment and medium. Background technique [0002] As an important function in the field of computational vision, target detection has extensive research value and application value. With the development of deep learning, the performance of target detection has been greatly improved, and it has been able to meet the needs of most scenarios. In the application process, there are often cases where the detection rate of objects of a certain scale is high and the detection rate of objects of other scales is low, and it cannot be guaranteed that objects of any scale can be effectively detected. In order to solve this problem, existing target detection methods consider introducing the same semantic features for models of objects of different scales or designing complex loss functions to focus on obje...

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): G06K9/62G06K9/00
CPCG06V20/40G06F18/29G06F18/214
Inventor 董健李帅丁明旭
Owner REMO 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