Target object detection method and device based on instance segmentation framework

A technology of target objects and frames, applied in the field of deep learning, can solve the problems of incomplete and inaccurate detection results, 2D position information cannot comprehensively and accurately reflect product feature information, etc., and achieve the effect of low cost.

Active Publication Date: 2021-07-27
GOERTEK INC
View PDF7 Cites 0 Cited by
  • Summary
  • Abstract
  • Description
  • Claims
  • Application Information

AI Technical Summary

Problems solved by technology

[0004] However, the 2D position information of the product often cannot fully and accurately reflect the feature information of the product, thus making the detection results incomplete and inaccurate

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 object detection method and device based on instance segmentation framework
  • Target object detection method and device based on instance segmentation framework
  • Target object detection method and device based on instance segmentation framework

Examples

Experimental program
Comparison scheme
Effect test

Embodiment 1

[0059] figure 1 is a block diagram of the hardware configuration of the detection system 100 according to the embodiment of the present invention.

[0060] Such as figure 1 As shown, the detection system 100 includes an image acquisition device 1000 and a target object detection device 2000 based on an instance segmentation framework.

[0061] The image acquisition device 1000 is configured to acquire a 2D image to be detected, and provide the acquired 2D image to be detected to the detection device 2000 .

[0062] The image acquisition device 1000 may be any imaging device capable of taking pictures of the object to be detected, such as a camera.

[0063] The detecting device 2000 may be any electronic equipment, such as a PC, a notebook computer, a server, and the like.

[0064] In this example, refer to figure 1 As shown, the detection device 2000 may include a processor 2100, a memory 2200, an interface device 2300, a communication device 2400, a display device 2500, a...

Embodiment 2

[0071] This embodiment provides a method for detecting a target object based on an instance segmentation framework, the method is as follows figure 2 As shown, including the following S201-S204:

[0072] S201. After starting target detection, acquire a 2D (Dimensional) image to be detected related to the target object.

[0073] In this embodiment, the target object refers to a product whose qualification needs to be determined, or a component in the product whose qualification needs to be determined. For example, the target object may be a patch capacitor on a PCB (Printed Circuit Board).

[0074] The 2D image to be detected is a 2D image obtained by taking pictures of the target object to be detected. The 2D image to be detected is related to the target object. That is to say, the target object is included in the 2D image to be detected. Of course, the target object may not be included in the 2D image to be detected.

[0075] S202. Input the 2D image to be detected into...

Embodiment 3

[0084] In this embodiment, the 3D position information of the target object obtained in S202 is represented by three rotation angles and three position offsets in three-dimensional space relative to the reference position. Based on this, the acquisition of the 3D position information of the target object in the above S202 includes the following steps S2021:

[0085] Obtain three rotation angles and position offsets of the target object relative to the reference position in three-dimensional space, and form a four-dimensional space vector as the 3D position information of the target object.

[0086] In this embodiment, the reference position is generally the position of the upper left corner of the 2D image to be detected.

[0087] In one example, a three-dimensional space can be represented by a three-dimensional coordinate system. The origin of the three-dimensional coordinate system may be the above-mentioned reference position, that is, the position of the upper left corne...

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 method and device for detecting a target object based on an instance segmentation framework. The method includes: after starting target detection, acquiring a 2D image to be detected related to the target object; inputting the 2D image to be detected into the instance segmentation framework, Use each network in this framework to process the 2D image to be detected. When running to the full convolutional network FCN of the instance segmentation framework, obtain the 3D position information of the target object, which indicates the rotation angle and position offset of the target object. ; Set the 3D weight factor corresponding to the 3D position information for the network loss function of the instance segmentation framework; use the 3D position information of the target object and the network loss function including the 3D weight factor to perform iterative calculations, and judge whether the target object is qualified according to the calculation results.

Description

technical field [0001] The present invention relates to the field of deep learning, in particular to a method for detecting a target object based on an instance segmentation framework, and a device for detecting a target object based on an instance segmentation framework. Background technique [0002] In the field of manufacturing and processing, it is usually necessary to check whether the produced products are qualified. [0003] At present, when testing whether a product is qualified, artificial intelligence AI technology is usually used to obtain the 2D position information of the product from the 2D image of the product, and then check whether the product is qualified according to the product feature information reflected in the 2D position information. [0004] However, the 2D position information of the product often cannot fully and accurately reflect the feature information of the product, thus making the detection result incomplete and inaccurate. [0005] Therefo...

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 Patents(China)
IPC IPC(8): G06T7/00G06T7/70G06K9/00G06K9/32G06N3/04
CPCG06T7/0004G06T7/70G06V20/10G06V10/25G06N3/045
Inventor 高巍张一凡于瑞涛
Owner GOERTEK INC
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