Check patentability & draft patents in minutes with Patsnap Eureka AI!

Power transmission line target detection method based on weighted deconvolution layer number improved DSSD algorithm, equipment and storage medium

A target detection and deconvolution technology, applied in the field of target detection, can solve the problems of reduced semantic information and low accuracy of small target detection, and achieve the effect of increasing mAP value, improving recognition accuracy, and improving accuracy

Active Publication Date: 2021-09-07
SHANDONG UNIV
View PDF9 Cites 1 Cited by
  • Summary
  • Abstract
  • Description
  • Claims
  • Application Information

AI Technical Summary

Problems solved by technology

At the same time, the detailed features of the feature map will reduce the semantic information as the depth increases during the network extraction process.
This results in a lower accuracy rate for small target detection

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
  • Power transmission line target detection method based on weighted deconvolution layer number improved DSSD algorithm, equipment and storage medium
  • Power transmission line target detection method based on weighted deconvolution layer number improved DSSD algorithm, equipment and storage medium
  • Power transmission line target detection method based on weighted deconvolution layer number improved DSSD algorithm, equipment and storage medium

Examples

Experimental program
Comparison scheme
Effect test

Embodiment 1

[0069] A power line target detection method based on the weighted deconvolution layer number improved DSSD algorithm, comprising the following steps:

[0070] (1) According to the existing power line identification technology based on the free algorithm of edge drawing parameters, the power line is identified; the image in step (1) is taken by the monitoring of the power transmission network, and the shooting angle is mostly overhead shooting, and the shooting range is ultra-wide-angle. 180° coverage, monitoring range 500m. The direction of the approximately trapezoidal transmission line is identified, and this process can be performed with reference to the prior art.

[0071] (2) First find the best fitting straight line for the two outermost edge transmission lines among the multiple transmission lines identified in step (1), and then take the best fitting trapezoid according to the area where the transmission line is located in the image, and keep the best fitting line. Th...

Embodiment 2

[0077] According to a kind of transmission line target detection method based on weighted deconvolution layer number improved DSSD algorithm described in embodiment 1, its difference is:

[0078] The network structure of the existing DSSD network model is as follows image 3 shown. The improved DSSD network model includes convolutional layer conv1, convolutional layer conv2_x, convolutional layer conv3_x, convolutional layer conv4_x, convolutional layer conv5_x, convolutional layer conv6_x, convolutional layer conv7_x, convolutional layer conv8_x, convolutional layer conv9_x, convolutional layer conv10_x, deconv1_x, deconv2_x, deconv3_x, deconv4_x, deconv5_x, deconv6_x;

[0079] Use an asymmetric feature pyramid structure for detection, such as Figure 4 As shown, after six layers of deconvolution layers, from the convolutional layer conv2_x, convolutional layer conv3_x, convolutional layer conv6_x, convolutional layer conv7_x, convolutional layer conv8_x, convolutional laye...

Embodiment 3

[0090] According to a kind of transmission line target detection method based on weighted deconvolution layer number improved DSSD algorithm described in embodiment 2, its difference is:

[0091] In step (4), the overall target loss function of the detection framework is determined by the center position loss L loc and the confidence loss L conf The weighted sum representation of , namely: the loss function is the weighted sum of the location error (locationloss, loc) and the confidence error (confidenceloss, conf): as shown in formula (II):

[0092]

[0093] In formula (II), N is the number of positive samples of the prior frame, c is the category confidence prediction value, I is the position prediction value of the corresponding bounding box of the prior frame, g is the position parameter of the groundtruth, and the weight coefficient α is passed cross-validation set to 1; middle Equal to 1 means that the i-th prior frame matches the j-th groundtruth, and the ground...

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 relates to a power transmission line target detection method based on a weighted deconvolution layer number improved DSSD algorithm, equipment and a storage medium. The method comprises the following steps: (1) identifying a power transmission line; (2) obtaining an optimal fitting straight line for two outermost transmission lines, obtaining an optimal fitting trapezoid according to the area where the transmission lines of the image are located, expanding the optimal fitting trapezoid without changing the center of gravity, wherein the expanded optimal fitting trapezoid is a detection range extracted from the input image; (3) constructing an improved DSSD network model; (4) training the improved DSSD network model; and (5) processing a to-be-detected image, and inputting the processed to-be-detected image into the trained and improved DSSD network model to obtain a detection result of the power transmission line. According to the method, targeted monitoring is carried out on the power transmission network, region division is completed according to the relation between the trend of the power transmission line and the distance, then efficient and accurate recognition of the small target is achieved, and the speed is increased while the small target recognition accuracy is considered.

Description

technical field [0001] The invention relates to a transmission line target detection method, equipment and storage medium based on the improved DSSD algorithm of weighted deconvolution layers, and belongs to the technical field of target detection. Background technique [0002] Object detection technology is a computer technology related to computer vision and image processing. In recent years, due to the development of convolutional neural networks and the improvement of hardware computing power, object detection based on deep learning is developing rapidly. In order to improve the generalization performance of neural networks, neural networks with different structures such as ALexNet, VGG, GoogLetNet and ResNet are proposed. Convolutional neural network (CNN) is the most representative deep learning model. The target detection algorithm has also been upgraded from a two-stage algorithm to a one-stage algorithm, which greatly improves the network speed. The regression-bas...

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
CPCG06N3/082G06N3/047G06N3/045G06F18/2415Y04S10/50
Inventor 范继辉赵明悦周莉
Owner SHANDONG UNIV
Features
  • R&D
  • Intellectual Property
  • Life Sciences
  • Materials
  • Tech Scout
Why Patsnap Eureka
  • Unparalleled Data Quality
  • Higher Quality Content
  • 60% Fewer Hallucinations
Social media
Patsnap Eureka Blog
Learn More

Browse by: Latest US Patents, China's latest patents, Technical Efficacy Thesaurus, Application Domain, Technology Topic, Popular Technical Reports.

© 2025 PatSnap. All rights reserved.Legal|Privacy policy|Modern Slavery Act Transparency Statement|Sitemap|About US| Contact US: help@patsnap.com