YOLO v5-based attached marine organism type identification method

A recognition method and marine biology technology, applied in neural learning methods, character and pattern recognition, biological neural network models, etc., can solve the problems of missed detection and low recognition accuracy, achieve good detection results, improve detection performance, reduce The effect of missed detection risk

Pending Publication Date: 2021-11-23
CHINA NUCLEAR IND MAINTENANCE
View PDF2 Cites 5 Cited by
  • Summary
  • Abstract
  • Description
  • Claims
  • Application Information

AI Technical Summary

Problems solved by technology

Among them, the YOLO v5 algorithm of the convolutional neural network is used to identify and detect the species of attached marine organisms, but the current detection algorithm has the problems of low recognition accuracy and missed 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
  • YOLO v5-based attached marine organism type identification method
  • YOLO v5-based attached marine organism type identification method
  • YOLO v5-based attached marine organism type identification method

Examples

Experimental program
Comparison scheme
Effect test

Embodiment

[0029] A method for identifying species of attached sea organisms based on YOLO v5, comprising the following steps:

[0030] S1. Collect pictures of marine biological attachments to construct a data set, divide the constructed data set into a training set, a test set and a verification set, specifically, collect pictures of marine biological attachments, and convert the obtained image format into .jpg format, Use the labelImg software to label the types of pictures, and the output format is .xml file, which is made into a standard VOC dataset format, and then the constructed dataset is randomly divided into training set, test set and verification set at a ratio of 8:1:1. And mark the independent name tags of each marine organism to complete the construction of the marine attached organisms data set.

[0031] Specifically, in the process of data construction, you can also use the program to crawl pictures on the Internet and various public datasets, and then use the dataset exp...

experiment example

[0048] In this experiment example, pictures of four kinds of attached marine organisms, spiral branch worms, sea moons, barnacles, and mussels were collected, and the acquired data set pictures were flipped, image enhanced, etc., and the format of the obtained pictures was converted to .jpg format, and use the labelImg software to label the types of pictures, the output format is .xml file, and make a standard VOC data set format, and the constructed data set is randomly divided into training set, test set and verification at a ratio of 8:1:1 set.

[0049] Use the k-means clustering algorithm for clustering to obtain 12 target Anchor values, and modify the original Anchor box to the new Anchor box after clustering, which are Anchor boxes: [281, 448], [303, 294], [ 512, 528], [139, 232], [200, 333], [228, 187], [83, 123], [96, 189], [131, 141], [40, 58], [51, 98], [78, 74].

[0050] The improved YOLO v5 algorithm is trained under the pytorch framework, and the input image for...

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 YOLO v5-based attached marine organism type identification method, which specifically comprises the steps of collecting marine organism attachment pictures to construct a data set, and dividing the constructed data set into a training set, a test set and a verification set; adding 1*1 convolution kernels in front of the first 3*3 convolution layer and the second 3*3 convolution layer of the Backbone of the YOLO v5 algorithm; adding a Non-local attention mechanism module respectively in front of the third BottleneckCSP of the Backbone and in front of the fourth BottleneckCSP of the Backbone; using a CSPNet structure composed of a Bottleneck and a standard convolution layer in the Backbone, and adding a detection layer of Output; introducing a PANet structure in the Neck; and generating an improved YOLO v5 algorithm, and finally, performing training by utilizing a data set, so that the accuracy of identifying the types of the attached marine organisms is higher, a good detection effect on small targets is achieved, the missing detection risk is reduced, and the detection performance of the traditional YOLO v5 algorithm is improved.

Description

technical field [0001] The invention relates to the technical field of image target detection, in particular to a YOLO v5-based method for identifying species of attached marine organisms. Background technique [0002] The ocean occupies 71% of the earth's area, providing human beings with abundant resources and potential wealth. In terms of global GDP, the income from marine resources development is as high as 7 trillion US dollars per year, and 90% of international trade transportation is undertaken by ocean transportation. [0003] With the continuous development of the ocean by humans, some problems have also been exposed. In coastal power plants, ships, and buildings in seawater, most of the substrate surfaces will be adhered by marine organisms, which will increase flow resistance and form localized corrosion cells, which will accelerate the corrosion rate of the substrate and cause unnecessary property losses. For example, most coastal power plants use seawater as t...

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): G06K9/62G06K9/46G06N3/04G06N3/08
CPCG06N3/08G06N3/045G06F18/23213G06F18/253G06F18/214
Inventor 李墨黄宇刘秉政王金龙徐广胤吕昌旗曹生现赵波李浩瑜
Owner CHINA NUCLEAR IND MAINTENANCE
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