CN114782875A - Fish fine-grained information acquisition method based on fishway construction - Google Patents
Fish fine-grained information acquisition method based on fishway construction Download PDFInfo
- Publication number
- CN114782875A CN114782875A CN202210539051.6A CN202210539051A CN114782875A CN 114782875 A CN114782875 A CN 114782875A CN 202210539051 A CN202210539051 A CN 202210539051A CN 114782875 A CN114782875 A CN 114782875A
- Authority
- CN
- China
- Prior art keywords
- fish
- image
- information
- fishway
- picture
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Granted
Links
- 238000000034 method Methods 0.000 title claims abstract description 14
- 238000010276 construction Methods 0.000 title claims abstract description 9
- 241000251468 Actinopterygii Species 0.000 claims abstract description 80
- 238000013508 migration Methods 0.000 claims abstract description 20
- 230000005012 migration Effects 0.000 claims abstract description 20
- 238000001514 detection method Methods 0.000 claims abstract description 11
- 238000013135 deep learning Methods 0.000 claims abstract description 9
- 230000000694 effects Effects 0.000 claims abstract description 9
- 238000013527 convolutional neural network Methods 0.000 claims description 23
- 238000012549 training Methods 0.000 claims description 13
- 238000013528 artificial neural network Methods 0.000 claims description 11
- 238000001914 filtration Methods 0.000 claims description 11
- 238000012545 processing Methods 0.000 claims description 6
- 241000894007 species Species 0.000 claims description 4
- 230000002708 enhancing effect Effects 0.000 claims description 3
- 238000006386 neutralization reaction Methods 0.000 claims description 2
- 238000013139 quantization Methods 0.000 claims description 2
- 238000010586 diagram Methods 0.000 abstract description 6
- 238000000605 extraction Methods 0.000 abstract description 4
- 238000012795 verification Methods 0.000 abstract description 4
- 238000000926 separation method Methods 0.000 abstract description 2
- 238000012360 testing method Methods 0.000 description 5
- 238000012937 correction Methods 0.000 description 4
- 230000001965 increasing effect Effects 0.000 description 3
- 238000007781 pre-processing Methods 0.000 description 3
- 238000009825 accumulation Methods 0.000 description 2
- 238000005516 engineering process Methods 0.000 description 2
- 230000006870 function Effects 0.000 description 2
- 238000013507 mapping Methods 0.000 description 2
- 230000008569 process Effects 0.000 description 2
- 238000013519 translation Methods 0.000 description 2
- 230000007306 turnover Effects 0.000 description 2
- XLYOFNOQVPJJNP-UHFFFAOYSA-N water Substances O XLYOFNOQVPJJNP-UHFFFAOYSA-N 0.000 description 2
- 241001270131 Agaricus moelleri Species 0.000 description 1
- 230000006399 behavior Effects 0.000 description 1
- 230000009286 beneficial effect Effects 0.000 description 1
- 230000008901 benefit Effects 0.000 description 1
- 238000009395 breeding Methods 0.000 description 1
- 230000001488 breeding effect Effects 0.000 description 1
- 238000011161 development Methods 0.000 description 1
- 239000012535 impurity Substances 0.000 description 1
- 239000011159 matrix material Substances 0.000 description 1
- 230000001617 migratory effect Effects 0.000 description 1
- 238000002156 mixing Methods 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- 230000008520 organization Effects 0.000 description 1
- 238000004321 preservation Methods 0.000 description 1
- OLBCVFGFOZPWHH-UHFFFAOYSA-N propofol Chemical compound CC(C)C1=CC=CC(C(C)C)=C1O OLBCVFGFOZPWHH-UHFFFAOYSA-N 0.000 description 1
- 229960004134 propofol Drugs 0.000 description 1
- 230000000246 remedial effect Effects 0.000 description 1
- 238000011160 research Methods 0.000 description 1
- 238000005070 sampling Methods 0.000 description 1
- 230000009182 swimming Effects 0.000 description 1
- 230000009466 transformation Effects 0.000 description 1
- 230000001131 transforming effect Effects 0.000 description 1
Images
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/24—Classification techniques
- G06F18/241—Classification techniques relating to the classification model, e.g. parametric or non-parametric approaches
- G06F18/2415—Classification techniques relating to the classification model, e.g. parametric or non-parametric approaches based on parametric or probabilistic models, e.g. based on likelihood ratio or false acceptance rate versus a false rejection rate
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/04—Architecture, e.g. interconnection topology
- G06N3/045—Combinations of networks
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/08—Learning methods
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T3/00—Geometric image transformations in the plane of the image
- G06T3/40—Scaling of whole images or parts thereof, e.g. expanding or contracting
- G06T3/4038—Image mosaicing, e.g. composing plane images from plane sub-images
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T2200/00—Indexing scheme for image data processing or generation, in general
- G06T2200/32—Indexing scheme for image data processing or generation, in general involving image mosaicing
-
- Y—GENERAL TAGGING OF NEW TECHNOLOGICAL DEVELOPMENTS; GENERAL TAGGING OF CROSS-SECTIONAL TECHNOLOGIES SPANNING OVER SEVERAL SECTIONS OF THE IPC; TECHNICAL SUBJECTS COVERED BY FORMER USPC CROSS-REFERENCE ART COLLECTIONS [XRACs] AND DIGESTS
- Y02—TECHNOLOGIES OR APPLICATIONS FOR MITIGATION OR ADAPTATION AGAINST CLIMATE CHANGE
- Y02A—TECHNOLOGIES FOR ADAPTATION TO CLIMATE CHANGE
- Y02A40/00—Adaptation technologies in agriculture, forestry, livestock or agroalimentary production
- Y02A40/60—Ecological corridors or buffer zones
Landscapes
- Engineering & Computer Science (AREA)
- Physics & Mathematics (AREA)
- Theoretical Computer Science (AREA)
- Data Mining & Analysis (AREA)
- General Physics & Mathematics (AREA)
- Evolutionary Computation (AREA)
- Life Sciences & Earth Sciences (AREA)
- Artificial Intelligence (AREA)
- General Engineering & Computer Science (AREA)
- General Health & Medical Sciences (AREA)
- Molecular Biology (AREA)
- Software Systems (AREA)
- Mathematical Physics (AREA)
- Computing Systems (AREA)
- Health & Medical Sciences (AREA)
- Biomedical Technology (AREA)
- Biophysics (AREA)
- Computational Linguistics (AREA)
- Probability & Statistics with Applications (AREA)
- Evolutionary Biology (AREA)
- Bioinformatics & Cheminformatics (AREA)
- Bioinformatics & Computational Biology (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Image Analysis (AREA)
- Image Processing (AREA)
Abstract
The invention discloses a fish fine-grained information acquisition method based on fishway construction, which is used for acquiring accurate information of multiple dimensions of fish based on the assistance of an underwater camera and a parallel grating, and improving the richness of image information and the robustness by utilizing image enhancement; the information separation degree of the target image is improved based on the algorithm of the edge preserving filter, and the accuracy of determining the contour and the position by the deep learning algorithm is improved; the accuracy of the model is further improved by utilizing a target detection algorithm Faster R-CNN based on deep learning; obtaining the schematic diagram of the activity cycle, the size and the condition of the body, the migration time and the migration path of each fish by using the position and the type information of the fish obtained by target detection; the parallel grating is used for carrying out secondary verification on the system precision, the precision of fish information extraction is greatly improved under the condition that the camera is mainly assisted by the parallel grating, and meanwhile, the living environment of fish is protected.
Description
Technical Field
The invention belongs to the technical field of image processing, and particularly relates to a fish fine-grained information acquisition method based on fishway construction.
Background
With the development of society, people are more and more concerned about the influence of the environment on human life. In many areas, the influence of the project on the local ecology is considered, and the local government also requires the organization of the project to submit a guarantee that the ecological environment is not affected during construction. When a hydraulic engineering is built, the influence of the engineering on the fishes in the area is usually detected to determine whether the engineering is feasible. Generally, pictures shot in water are high in turbidity degree and dark in light and have interference of other impurities, and in addition, the robustness of the used algorithm is poor, so that complete and detailed accurate information of the fishes is difficult to extract. Therefore, under the condition of certain turbidity and light interference, the technology for extracting the relevant information of the fish through fishway construction is urgent.
The fishway is a passage for fish migration in hydraulic engineering, and the behavior of the fish is generally protected by building artificial water tanks on sluice gates and dams as human activities destroy the passage for fish migration and take remedial measures. Generally, an underwater camera is installed in a fishway to obtain migration information of fishes, and the migration information is used for analyzing the breeding condition and habit of the fishes. At present, a small amount of information of fishes is obtained by using an underwater camera, but the obtaining difficulty is high, the obtained images are not clear, a physical framework of a system is not provided for obtaining corresponding information, and the local ecological environment is possibly damaged.
Disclosure of Invention
In order to solve the problems in the prior art, the invention provides a fish fine-grained information acquisition method based on fishway construction.
The specific technical scheme of the invention is as follows: a fish fine-grained information acquisition method based on fishway construction comprises the following steps:
s0, installing a group of parallel gratings on the front and back of the fishway, respectively installing an underwater camera at the inlet and outlet of the front of the fishway, wherein the parallel gratings are used for recording pixel map information of the fish, and the underwater cameras are used for recording image information of the fish when the fish enters and exits the fishway;
s1, obtaining a fish migration video, extracting a feature picture, and enhancing the obtained feature picture;
s2, carrying out filtering preprocessing on the picture enhanced in the step S1 to obtain a filtered data set;
s3, training the data set obtained in the step S2 by selecting an Faster R-CNN target detection algorithm based on deep learning to obtain a deep neural network;
s4, obtaining an activity cycle, a size, a migration time and a migration path under the condition that the fish species and the fish positions are known;
s5, extracting pixel maps of parallel rasters according to frames, putting the obtained group of brand-new pixel maps as a second training set into a Faster R-CNN convolutional neural network in the step S3 for training to obtain a group of new weight parameters, putting the same picture in the step S4 into a new neural network for verifying accuracy, outputting information in the step S4 if the two results are the same, and re-performing the steps S1-S4 if the two results are different.
In step S1, a long video with migratory fish passing through is captured from the underwater camera in the fishway, and feature pictures are manually selected or pictures are extracted frame by frame at a certain time interval. The method comprises the steps of correcting and transforming gray scales or modifying a histogram by using an image enhancement algorithm based on a spatial domain, and increasing the information content of a data set by simply carrying out operations such as scaling, translation and turnover on fish contours in an image, so as to expand the data set and increase the target identification capability of a deep neural network.
In step S2, edge preservation is performed on the picture enhanced in step S1And filtering, wherein in the filtering process, the brightness sensed by human eye cells and the image brightness are considered to form a Log relation, and the brightness is converted into a Log domain and then normalized: l ═ ln (L)in·106+1), wherein L and LinThe values of luminance in the Log domain and the actual image luminance are respectively.
Specifically, in step S2, the picture enhanced in step S1 is subjected to three-layer local edge preserving filtering processing, and finally three-layer images are weighted to obtain a final output image, that is, an image processed by the edge preserving filter, wherein in order to set the brightness mapped in the Log domain to be between 0 and 1, the processed image is subjected to neutralization through two LEP filters, and then the obtained three results are subjected to Lout=D1′·0.5+D2′+D3' the weighted sum is linearly mapped to the interval of 0 to 1, where D is the final image1' is the difference between the image subjected to the LEP filter twice and the image after the quantization, D2' is the difference between the image passed through the two LEP filters and the image passed through the one LEP filter, D3' is the difference between the image that has passed through the LEP filter once and the original image.
In step S4, the average speed of the fish moving in the fishway is obtained by dividing the distance by the time according to the observed species and position information of the fish moving in a certain time; the length and the width of the fish in the fishway can be obtained by proportion according to the principle of similar triangles (under the condition that the distance between the reference object and the camera and the length of the reference object are known); calculating the time for the fish to pass through the fishway by calculating the time difference of shooting the same fish at the fishway opening; simulating the position of the fish in the target detection as a point (the center of the rectangle) can splice each picture in the video frame by frame to obtain a migration route in the fishway.
The invention has the beneficial effects that: the fish fine-grained information acquisition method is based on a brand-new fishway framework, namely an underwater camera and a parallel grating auxiliary method, systematically acquires accurate information of multiple dimensions of fish, and improves the richness of image information by utilizing image enhancement, thereby improving the robustness of the method; the information separation degree of the target image is improved based on the algorithm of the edge-preserving filter in the nonlinear filter, and the accuracy of determining the contour and the position by the depth learning algorithm is improved; the accuracy of the model is further improved by using a target detection algorithm, namely, Faster R-CNN based on deep learning; obtaining the schematic diagram of the activity cycle, the size and the condition of the body, the migration time and the migration path of each fish by using the position and the type information of the fish obtained by target detection; the parallel grating is used for carrying out secondary verification on the system precision, the precision of fish information extraction is greatly improved under the condition that the camera is mainly used for assisting the parallel grating, and meanwhile, the living environment of fish is protected.
Drawings
FIG. 1 is a schematic flow chart of the steps performed in the present invention.
Fig. 2 is a perspective schematic view of a fishway in an embodiment of the invention.
FIG. 3 is a diagram illustrating a layer of processing in an edge preserving filter according to an embodiment of the present invention.
FIG. 4 is a structural diagram of the Faster R-CNN algorithm based on the deep learning neural network in the embodiment of the present invention.
Detailed Description
The technical scheme of the invention is further explained by combining the attached drawings.
The fish pictures mentioned in this embodiment all come from an underwater camera and a group of underwater parallel gratings in the fishway, and have certain turbidity and light interference. In the conventional fish information extraction, because the underwater shooting difficulty is high, a large enough number of pictures can not be obtained as a training set to obtain accurate fish parameters, and the problem is solved by image enhancement. The image enhancement is to purposefully emphasize the whole or local features of an image, to make an unclear image clear or emphasize some interesting features, to enlarge the difference of different objects between images and to inhibit the uninteresting features, so as to improve the quality of the image, to enrich the information content, and to have good effect after being used in a later deep learning training session. When the fish contour is identified, the edge contour of the fish can be highlighted by using a sharpening operation based on a domain enhancement algorithm, so that the target detection accuracy is improved. For the target picture, the contour and position information of the fish can be strengthened through image filtering. The specific process is shown in fig. 1, and comprises the following steps:
s0, installing the parallel grating and two underwater cameras according to the pattern shown in fig. 2. The parallel gratings are positioned in the center of the front and back of the fishway and parallel to each other, and the underwater camera is positioned at the midpoint of the front of the inlet and outlet of the fishway. The fish enter the left from the right and exit, the two semicircles are two cameras at the entrance and exit, and the two cylinders are a set of parallel gratings.
And S1, obtaining the fish migration video, extracting the characteristic picture, and enhancing the obtained characteristic picture.
Under the condition that the data set is not large enough, the spatial domain-based image enhancement algorithm is used for carrying out gray correction and transformation or histogram correction, and in addition, the data set information content is increased through simple operations of scaling, translation, turnover and the like of fish contours in the image, so that the robustness of the model is improved. When video information is extracted frame by frame, attention is paid to videos containing different seasons, different time periods, and different weather. Each picture in the obtained data set is numbered according to 1, 2 and 3 … …, and each picture is subjected to gray correction (brightness of each pixel on the image is adjusted) to obtain a new picture and the new picture is stored in the current directory. And (3) performing histogram correction on the pictures in the data set, firstly calculating the probability of each gray level, calculating the gray level accumulation probability, then multiplying the accumulation probability by length-1, then rounding, updating the gray level, obtaining a new picture, and storing the new picture in the current directory. And translating and turning the position of the framed fish, or scaling the size of the fish according to the proportion to obtain new images, and storing the new images in the current directory. Sharpening operations (high-pass filtering, operators and the like) based on a domain enhancement algorithm can be added to highlight the edge contour of the fish, and the new images are also stored in the current directory to obtain a brand new data set which contains various abundant information of the fish and is used for training the following neural network.
And S2, carrying out filtering pretreatment on the picture enhanced in the step S1 to obtain a filtered data set.
FIG. 3 is a diagram illustrating a layer of processing in the edge preserving filter of the present invention. Firstly, preprocessing brightness, and obtaining the brightness by the formula L ═ ln (L)in·106+1), corresponding the brightness of the pixel points in the image to the Log domain. Linear mapping of values in the Log domain to [0,1 ]]In this interval, the following filtering operation is performed, in which an original unfiltered image is set as a, a B image is obtained after passing through a local edge preserving filter (LEP) filter once, a C image is obtained after passing through an LEP filter once, and the C image is averaged, that is, an average (mean) step in the image, to obtain a D image. The image D and the image C are subjected to difference to obtain a D1 image; the image C and the image B are subjected to difference to obtain a D2 image; and subtracting the image B from the image A to obtain a D3 image, and performing weighting operation, wherein the weight is D1, D2, D3 is 0.5:1:1, namely the formula: l is a radical of an alcoholout=D1′·0.5+D2′+D3', a pre-output image is obtained. Linearly projecting the pixel point of the image to [0,1 ]]A true output image is obtained in this region.
S3, training the data set by selecting the fast R-CNN target detection algorithm based on deep learning to obtain a deep learning neural network, as shown in figure 4. Compared with a Fast R-CNN algorithm, the Fast R-CNN algorithm adopts RPN (region pro positive network) to replace an advice window generated by an original Selective Search, and CNN of the advice window can be shared with CNN of target detection.
And after the images in the data set are subjected to data enhancement, preprocessing is carried out on the target image, an edge preserving filter is used for reducing the noise interference, the position and outline information of the fish are kept, and the fish are placed in a trained fast RCNN neural network to obtain the species and position information of the fish. In a structural schematic diagram of a Faster R-CNN algorithm, firstly, a test picture is input, the whole picture is input into the CNN for feature extraction, a pile of anchor frames (anchor box) is generated by RPN, the anchor frames are cut and filtered, and then whether anchor points (anchors) belong to a foreground or a background, namely an interest target or a background, is judged through SoftMax, wherein the anchor points refer to that in a target detection task, an input image is extracted through a backbone network to obtain a feature map, and each pixel point on the map is the anchor point. And the other frame regression (bounding box regression) corrects the anchor frame to form a more accurate candidate region (propofol). And mapping the suggestion window onto the last layer of feature map (feature map) of the CNN, and performing combined training on the classification profile and the frame regression by using a SoftMax loss function and a smooth loss function, wherein the obtained weight is a required parameter of the convolutional neural network.
And S4, under the condition that the type and the position of the fish are known, obtaining information such as activity cycle, body size, migration time, migration path and the like through mathematical processing to extract underwater fish information.
As shown in fig. 2, an underwater high-definition camera is arranged at the top end of the fishway, an underwater high-definition camera is also arranged at the tail end of the fishway, video information of passing fishes in a long period of time can be recorded, and a pair of infrared gratings is arranged right above the fishway. After the neural network is trained in the previous step, the target picture processed in step S2 is put into a convolutional neural network, and the position coordinates, length and width of the central point of the fish in the picture and the species information of the fish are obtained. The ir grating can display the specific length of the fish in its corresponding application program by a similar ratio a: a ═ B: B, where a is the specific length of the fish, B is the actual width of the fish, a is the pixel length (several pixels) corresponding to the fish in the convolutional neural network, and B is the pixel width (several pixels) corresponding to the fish in the convolutional neural network. Therefore, b can be solved, and the real data of the body size can be obtained.
By recording the time interval during which the number of swimming fish is increased dramatically, the activity cycle of the fish can be obtained. And the migration time needs to be marked out by using corresponding labels in the convolutional neural network, and the time difference of the fishes passing through the two underwater high-definition cameras is the migration time. Knowing the position coordinates of the central point of the fish in each picture, splicing each picture passing through the convolutional neural network by using an image splicing technology, specifically extracting the features of the pictures by using an SIFT algorithm, matching the features of the pictures by using a random sampling consistency algorithm, and finally performing image splicing (Blending) by using matching key points to obtain the motion path of each fish in the fishway, so that the ecological research of the fish is facilitated.
And S5, recording the fish outline information by using a group of underwater parallel gratings so as to facilitate the second identification of the fish type and improve the accuracy of the algorithm. The grating can effectively record the outline and position information of an object, and the obtained grating image is used as a second training set to check the fish species information.
Obtaining a series of time-varying dot-matrix graphs from the group of underwater rasters, truncating the pictures frame by frame to obtain images with fish outlines to form a training set containing a brand-new pixel graph, then repeating the steps S1-S4 to obtain a brand-new neural network parameter set, and putting the pictures for testing in S4 into a new convolutional neural network for testing. If the output result is the same as the previous output result, the result of the secondary verification is correct, the result is really a trustworthy result, and the result is presented as the final output of the system. If the output result is different from the previous output result, the secondary verification is not established and is not a reliable result, and a new test picture is reselected for testing so as to improve the accuracy of the whole system.
It will be appreciated by those of ordinary skill in the art that the embodiments described herein are intended to assist the reader in understanding the principles of the invention and are to be construed as being without limitation to such specifically recited embodiments and examples. Those skilled in the art, having the benefit of this disclosure, may effect numerous modifications thereto and changes may be made without departing from the scope of the invention in its aspects.
Claims (2)
1. A fish fine-grained information acquisition method based on fishway construction comprises the following steps:
s0, installing a group of parallel gratings on the front and back of the fishway, respectively installing an underwater camera at the inlet and outlet of the front of the fishway, wherein the parallel gratings are used for recording pixel map information of the fish, and the underwater cameras are used for recording image information of the fish when the fish enters and exits the fishway;
s1, obtaining a fish migration video, extracting a feature picture, and enhancing the obtained feature picture;
s2, carrying out filtering pretreatment on the picture enhanced in the step S1 to obtain a filtered data set;
s3, training the data set obtained in the step S2 by selecting a fast R-CNN target detection algorithm based on deep learning to obtain a deep neural network;
s4, obtaining activity cycle, size, migration time and migration path under the condition that the species and the position of the fishes are known;
s5, extracting pixel maps of parallel rasters according to frames, putting the obtained group of brand-new pixel maps as a second training set into a Faster R-CNN convolutional neural network in the step S3 for training to obtain a group of new weight parameters, putting the same picture in the step S4 into a new neural network for verifying accuracy, outputting information in the step S4 if the two results are the same, and re-performing the steps S1-S4 if the two results are different.
2. The method as claimed in claim 1, wherein in step S2, the picture enhanced in step S1 is subjected to three-layer local edge preserving filtering processing, and the three-layer image is finally weighted to obtain a final output image, i.e., an image processed by the edge preserving filter, wherein in order to set the luminance mapped in the Log domain to be between 0 and 1, the processed image is subjected to two times of local edge preserving filter (LEP) filtering, the three results are subjected to neutralization, and the three results are obtained according to Lout=D1′·0.5+D2′+D3' the weighted sum is linearly mapped to the interval of 0 to 1, where D is the final image1' is passed through two LEP filtersDifference between image and image after quantization, D2' is the difference between the image passed through the two LEP filters and the image passed through the one LEP filter, D3' is the difference between the image that has passed through the LEP filter once and the original image.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN202210539051.6A CN114782875B (en) | 2022-05-18 | 2022-05-18 | Fish fine granularity information acquisition method based on fishway construction |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN202210539051.6A CN114782875B (en) | 2022-05-18 | 2022-05-18 | Fish fine granularity information acquisition method based on fishway construction |
Publications (2)
Publication Number | Publication Date |
---|---|
CN114782875A true CN114782875A (en) | 2022-07-22 |
CN114782875B CN114782875B (en) | 2023-04-21 |
Family
ID=82437122
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN202210539051.6A Active CN114782875B (en) | 2022-05-18 | 2022-05-18 | Fish fine granularity information acquisition method based on fishway construction |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN114782875B (en) |
Cited By (1)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN116363494A (en) * | 2023-05-31 | 2023-06-30 | 睿克环境科技(中国)有限公司 | Fish quantity monitoring and migration tracking method and system |
Citations (4)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
US20190241633A1 (en) * | 2016-05-04 | 2019-08-08 | Curevac Ag | Rna encoding a therapeutic protein |
CN111612017A (en) * | 2020-07-07 | 2020-09-01 | 中国人民解放军国防科技大学 | Target detection method based on information enhancement |
CN112926652A (en) * | 2021-02-25 | 2021-06-08 | 青岛科技大学 | Fish fine-grained image identification method based on deep learning |
CN114766034A (en) * | 2020-06-03 | 2022-07-19 | 辉达公司 | Object detection system based on machine learning |
-
2022
- 2022-05-18 CN CN202210539051.6A patent/CN114782875B/en active Active
Patent Citations (4)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
US20190241633A1 (en) * | 2016-05-04 | 2019-08-08 | Curevac Ag | Rna encoding a therapeutic protein |
CN114766034A (en) * | 2020-06-03 | 2022-07-19 | 辉达公司 | Object detection system based on machine learning |
CN111612017A (en) * | 2020-07-07 | 2020-09-01 | 中国人民解放军国防科技大学 | Target detection method based on information enhancement |
CN112926652A (en) * | 2021-02-25 | 2021-06-08 | 青岛科技大学 | Fish fine-grained image identification method based on deep learning |
Non-Patent Citations (3)
Title |
---|
CHRISTIAN M. D. TRINH 等: "\"Biometric Fish Classification of Nordic Species Using Convolutional Neural Network with Squeeze-and-Excitation\"" * |
JUNHAI LUO 等: "Improved Cubature Kalman Filter for Target Tracking in Underwater Wireless Sensor Networks" * |
范雕: "卫星测高重力数据反演海底地形的理论和方法研究" * |
Cited By (1)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN116363494A (en) * | 2023-05-31 | 2023-06-30 | 睿克环境科技(中国)有限公司 | Fish quantity monitoring and migration tracking method and system |
Also Published As
Publication number | Publication date |
---|---|
CN114782875B (en) | 2023-04-21 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
Jia et al. | Detection and segmentation of overlapped fruits based on optimized mask R-CNN application in apple harvesting robot | |
CN113065558B (en) | Lightweight small target detection method combined with attention mechanism | |
Terayama et al. | Integration of sonar and optical camera images using deep neural network for fish monitoring | |
CN111046880A (en) | Infrared target image segmentation method and system, electronic device and storage medium | |
CN109559302A (en) | Pipe video defect inspection method based on convolutional neural networks | |
CN110544251A (en) | Dam crack detection method based on multi-migration learning model fusion | |
CN101782526B (en) | Method and device for automatically restoring, measuring and classifying steel dimple images | |
CN113011357A (en) | Depth fake face video positioning method based on space-time fusion | |
CN109543691A (en) | Ponding recognition methods, device and storage medium | |
CN111339902B (en) | Liquid crystal display indication recognition method and device for digital display instrument | |
CN109635634A (en) | A kind of pedestrian based on stochastic linear interpolation identifies data enhancement methods again | |
CN111640138B (en) | Target tracking method, device, equipment and storage medium | |
CN107945210A (en) | Target tracking algorism based on deep learning and environment self-adaption | |
CN114724022A (en) | Culture fish school detection method, system and medium fusing SKNet and YOLOv5 | |
Mai et al. | Back propagation neural network dehazing | |
CN116229319A (en) | Multi-scale feature fusion class behavior detection method and system | |
CN110853041A (en) | Underwater pier component segmentation method based on deep learning and sonar imaging | |
CN114782875B (en) | Fish fine granularity information acquisition method based on fishway construction | |
CN113591592A (en) | Overwater target identification method and device, terminal equipment and storage medium | |
CN115578423A (en) | Fish key point detection, individual tracking and biomass estimation method and system based on deep learning | |
CN115546187A (en) | Agricultural pest and disease detection method and device based on YOLO v5 | |
CN110674823A (en) | Sample library construction method based on automatic identification of deep sea large benthonic animals | |
CN112906510A (en) | Fishery resource statistical method and system | |
CN117058232A (en) | Position detection method for fish target individuals in cultured fish shoal by improving YOLOv8 model | |
CN116612272A (en) | Intelligent digital detection system for image processing and detection method thereof |
Legal Events
Date | Code | Title | Description |
---|---|---|---|
PB01 | Publication | ||
PB01 | Publication | ||
SE01 | Entry into force of request for substantive examination | ||
SE01 | Entry into force of request for substantive examination | ||
GR01 | Patent grant | ||
GR01 | Patent grant |