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

Classification regression characteristic decoupling method in target detection

A target detection and feature map technology, applied in the field of target detection, can solve problems such as feature coupling, and achieve the effect of maintaining consistency and improving overall accuracy

Active Publication Date: 2020-11-06
SOUTH CHINA UNIV OF TECH
View PDF5 Cites 1 Cited by
  • Summary
  • Abstract
  • Description
  • Claims
  • Application Information

AI Technical Summary

Problems solved by technology

[0004] However, in the current mainstream target detection algorithms, such as Faster RCNN, there is still a phenomenon of feature coupling.

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
  • Classification regression characteristic decoupling method in target detection
  • Classification regression characteristic decoupling method in target detection
  • Classification regression characteristic decoupling method in target detection

Examples

Experimental program
Comparison scheme
Effect test

Embodiment Construction

[0040] The following will clearly and completely describe the technical solutions in the embodiments of the present invention with reference to the accompanying drawings in the embodiments of the present invention. Obviously, the described embodiments are only some, not all, embodiments of the present invention. All other embodiments obtained by persons of ordinary skill in the art based on the embodiments of the present invention belong to the protection scope of the present invention.

[0041] According to an embodiment of the present invention, a classification regression feature decoupling method in target detection is provided.

[0042] Such as figure 1 As shown, the classification regression feature decoupling method in the target detection according to the embodiment of the present invention includes the following steps:

[0043] Step S1, use the network structure of the regression part to correct the bounding box in advance, return the corrected information to the net...

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 classification regression characteristic decoupling method in target detection and relates to the technical field of target detection. The method comprises the following steps: correcting a bounding box by using a network structure of a regression part in advance, returning corrected information to a network, executing a network of a classification part, and carrying outcontinuous iteration to establish a relationship between a score of the classification part and the corrected bounding box; and decoupling the classification and regression network. According to the method, correction information of the boundary prediction frame is fed back by the regression network, so that the classification network can learn information returned by the regression network, and the consistency of the positioning confidence coefficient and the classification confidence coefficient of the boundary frame is kept.

Description

technical field [0001] The invention relates to the technical field of target detection, in particular to a method for decoupling classification and regression features in target detection. Background technique [0002] Object detection is mainly an important research direction in the field of computer vision, and it plays an indispensable role in many applications such as face detection and vehicle detection. With the development of deep learning technology, compared with traditional target detection methods, target detection algorithms based on deep learning have made great progress in algorithm accuracy. [0003] Compared with image recognition, target detection not only needs to distinguish the category of the target in the image, but also needs to regress the position of the target in the image. [0004] However, the phenomenon of feature coupling still exists in the current mainstream target detection algorithms, such as Faster RCNN. Specifically, the network sharing...

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): G06K9/00G06K9/62G06N3/04G06N3/08
CPCG06N3/08G06V20/41G06N3/045G06F18/214
Inventor 蒋馨瑶何盛烽
Owner SOUTH CHINA UNIV OF TECH