Intelligent fish sorting method and system based on deep feature fusion

A deep feature and sorting system technology, applied in neural learning methods, fish farming, instruments, etc., can solve the problems of easy death or injury of fish, high labor intensity, death or injury of fish, and avoid injury or damage. Effects of death, improved accuracy, and strong feature capture ability

Pending Publication Date: 2022-04-29
SOUTH CHINA AGRI UNIV
View PDF0 Cites 0 Cited by
  • Summary
  • Abstract
  • Description
  • Claims
  • Application Information

AI Technical Summary

Problems solved by technology

Manual identification and sorting has the following defects: 1. High labor intensity, low sorting efficiency, and high error rate; 2. Fish identification and sorting takes a long time to expose, and fish are prone to death or damage; 3. Due to the dependence on the experience of skilled workers , the level of skilled workers determines the correct rate of sorting, and the efficiency of identification and sorting is unstable
Although the fish automatic identification and sorting technology has also appeared at present, there are the following problems: (1) the identification and sorting is carried out in a non-aqueous environment, which is likely to cause fish death or damage; (2) the identification technology is relatively primitive, and the identification The accuracy rate is not enough; (3) Various propellers are often used for fish sorting, and the collision between the propeller and the fish body can easily cause damage to the fish body, affecting the survival rate and quality of the fish

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
  • Intelligent fish sorting method and system based on deep feature fusion
  • Intelligent fish sorting method and system based on deep feature fusion
  • Intelligent fish sorting method and system based on deep feature fusion

Examples

Experimental program
Comparison scheme
Effect test

Embodiment 1

[0046] An intelligent fish sorting system based on deep feature fusion of the present invention includes a fish storage tank 1 for containing fish to be identified, a valve 2, a delivery pipe 3, a fish identification module 4 and a collection module 5. The fish identification module 4 includes an identification pool 41 , an identifier 42 and a controller 43 for identifying fish types. One end of the valve 2 communicates with the bottom of the fish pond 1, and the other end communicates with one end of the delivery pipe 3. The conveying pipe 3 is a circular pipe, and the other end of the conveying pipe 3 communicates with the identification pool 41 . The height of fish storage pond 1 is higher than the height of identification pond 41. The pipe diameter of valve 2 and conveying pipe 3 only allows one fish to pass through, guarantees that after valve 2 is opened each time, only one fish enters conveying pipe 3 and identification pool 41 through valve 2 in fish storage pond 1 fo...

Embodiment 2

[0051] A kind of fish intelligent sorting method based on deep feature fusion of the present invention comprises the following:

[0052] (1) The camera 421 shoots the image of the fish to be identified in the identification pool 41, and sends the image of the fish to be identified to the artificial intelligence algorithm processor 422, and the artificial intelligence algorithm processor 422 obtains the image of the fish to be identified, and uses the image preprocessing positioning algorithm , preprocess the image, and perform real-time location of the fish to be identified. The image preprocessing positioning algorithm adopts the background difference method, and uses the Gaussian mixture model to update the background of the image in real time, including the following specific steps:

[0053] (1a) Obtain the first frame image captured by the camera 421 as the initialization background image;

[0054] (1b) Call the cv.createBackgroundSubtractorMOG2() function in the Opencv i...

Embodiment 3

[0089] A method for using a fish intelligent sorting system based on deep feature fusion of the present invention includes the following:

[0090] Step 1, putting the fish caught by the net into the fish storage pond, the mesh of the fishing net is relatively large, and during the fishing process, smaller fish are leaked from the mesh;

[0091] Step 2, the controller 43 activates the valve 2, the fish to be identified in the fish pond 1 swims into the delivery pipe 3 through the valve 2, and the valve 2 is closed. The fish to be identified swims into the identification pool 41 along the delivery pipe 3, and the camera 421 takes an image of the fish to be identified and sends it to the artificial intelligence algorithm processor 422;

[0092] Step 3, the artificial intelligence algorithm processor 422 receives the image of the fish to be identified, preprocesses the image, and locates the fish to be identified in real time, extracts features from the image through a deep learni...

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 intelligent fish sorting method based on deep feature fusion, and the method comprises the steps: (1), obtaining an image of a to-be-recognized fish, carrying out the preprocessing through an image preprocessing positioning algorithm, and carrying out the real-time positioning; (2) performing feature extraction on the preprocessed image through a deep learning algorithm, inputting the extracted features into a pre-trained convolutional neural network model, classifying the image to obtain the type of the to-be-identified fish, and outputting an identification result; and (3) the intelligent fish sorting system temporarily stores the to-be-recognized fishes according to the recognition result and the types, and sorting is completed. According to the method, the convolutional neural network based on deep feature fusion is adopted, an up-sampling layer is added to a deep convolutional layer and is spliced and fused with features of a second deconvolutional layer, deep edge detail information and shallow contour features of the image are combined, the method has higher feature capture capability, the features of the image are comprehensively mined, and the image quality is improved. And the fish identification accuracy in the water environment is improved.

Description

technical field [0001] The invention relates to the technical field of image recognition and mechanical device design, in particular to an intelligent fish sorting method and system based on deep feature fusion. Background technique [0002] In order to maximize the use of aquaculture water and improve aquaculture efficiency, multi-species polyculture is usually implemented when artificially aquaculture fish. There are also many kinds of fish in rivers and oceans. Therefore, fish sorting is involved no matter whether it is fishing in an artificial breeding environment or in a natural water body in the wild. At present, the identification and sorting of different fish species is mainly carried out by skilled workers. Manual identification and sorting has the following defects: 1. High labor intensity, low sorting efficiency, and high error rate; 2. Fish identification and sorting takes a long time to expose, and fish are prone to death or damage; 3. Due to the dependence on...

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): G06V10/764G06V10/80G06V10/82G06V10/44G06N3/04G06N3/08A01K61/95G06K9/62
CPCG06N3/084A01K61/95G06N3/045G06F18/24G06F18/253
Inventor 曾芳李紫聪卢俊菠曾君刘俊峰
Owner SOUTH CHINA AGRI UNIV
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