A method and system for the prevention and control of tuna blastula overflow and biological predation

Abstract

The application discloses a method and system for preventing and controlling tuna fertilized egg overflow and biological predation, comprising: obtaining underwater real-time monitoring information, performing target detection based on the underwater real-time monitoring information to obtain target detection information; identifying the detection target based on the target detection information to identify yellowfin tuna and fertilized eggs and obtain underwater target identification information; performing target tracking according to the underwater target identification information, analyzing the movement path of the yellowfin tuna and the fertilized eggs in a unit time to obtain movement path analysis information; constructing a water flow field simulation model, analyzing the water flow field condition in the target aquaculture net cage, and performing fertilized egg movement trajectory prediction to obtain fertilized egg movement trajectory prediction information; judging whether the target fertilized egg is at risk based on the fertilized egg movement trajectory prediction information, generating risk warning information and formulating a control scheme. The yield and quality of tuna culture are effectively improved.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of aquaculture, and in particular to a method and system for preventing and controlling overflow and biological predation of tuna fertilized eggs. BACKGROUND

[0002] As a high economic value marine species, the artificial breeding technology of yellowfin tuna has been widely used. However, during the net cage culture process, the overflow and predation of tuna fertilized eggs by other organisms has become an important technical problem. The overflow of fertilized eggs will lead to uncontrollable loss of tuna populations, not only affecting the economic benefits of cultivation, but also increasing the management difficulty of tuna populations in the environment. At the same time, due to the small size of the fertilized eggs, which float on the surface or middle layer of the water body, they are easily preyed upon by other organisms in the ocean, resulting in a decrease in the survival rate of the fertilized eggs. This problem limits the further development of tuna cultivation. Therefore, how to better monitor the fertilized eggs in the cultivation net cage to avoid overflow and biological predation is an important problem. SUMMARY

[0003] The present application overcomes the defects of the prior art and provides a method and system for preventing and controlling the overflow and biological predation of tuna fertilized eggs, which aims to improve the yield and quality of tuna cultivation.

[0004] To achieve the above-mentioned purpose, the first aspect of the present application provides a method for preventing and controlling the overflow and biological predation of tuna fertilized eggs, comprising:

[0005] Obtaining underwater real-time monitoring information of a target cultivation net cage, performing target detection based on the underwater real-time monitoring information, and obtaining target detection information;

[0006] Constructing an underwater target recognition model, recognizing the detection target based on the target detection information, recognizing yellowfin tuna and fertilized eggs, and obtaining underwater target recognition information;

[0007] Obtaining underwater target recognition information, performing target tracking according to the underwater target recognition information, analyzing the movement path of yellowfin tuna and fertilized eggs within a unit time, and obtaining movement path analysis information;

[0008] Constructing a water flow field simulation model, obtaining water flow velocity monitoring information, analyzing the water flow field conditions in the target cultivation net cage, and performing fertilized egg movement trajectory prediction, and obtaining fertilized egg movement trajectory prediction information;

[0009] Judging whether the target fertilized egg is at risk based on the fertilized egg movement trajectory prediction information, generating risk warning information, and formulating a control scheme.

[0010] In the scheme, the underwater real-time monitoring information of the target aquaculture net cage is obtained, target detection is performed based on the underwater real-time monitoring information, and target detection information is obtained, specifically including:

[0011] The target aquaculture net cage is monitored, underwater real-time monitoring information of the target aquaculture net cage is obtained based on underwater video monitoring technology, and image preprocessing is performed to obtain preprocessed image information;

[0012] A target detection model is constructed based on YOLOv5, the preprocessed image information is input for target detection, feature extraction is performed on the preprocessed image information, and an initial feature map is generated;

[0013] According to the initial feature map, the scale features corresponding to each feature are obtained, the attention mechanism is introduced to obtain the attention scores of each feature, the attention feature map is generated, and the initial feature map and the attention feature map are fused by using the feature fusion pyramid;

[0014] The candidate detection frame is generated according to the fused feature map, the non-maximum suppression method is used for screening, the final target detection frame is obtained, and the target detection information is obtained through the final target detection frame.

[0015] In the scheme, the underwater target recognition model is constructed, the detection target is recognized based on the target detection information, the yellowfin tuna and the fertilized egg are recognized, the underwater target recognition information is obtained, and specifically including:

[0016] Based on data retrieval, fish body images and fertilized egg images of underwater yellowfin tuna are obtained, the obtained image information is preprocessed, sample expansion processing is performed according to the preprocessed image, and an instance data set is formed;

[0017] An underwater target recognition model is constructed through a neural network, a training data set is constructed using the instance data set to perform deep learning and training on the underwater target recognition model;

[0018] The target detection information is obtained, the detection target image is obtained by using an image segmentation algorithm, and is input into a stacked denoising autoencoder for encoding to obtain a low-dimensional latent representation of the detection target image and construct a latent space;

[0019] The attribute combination of the detection target image is obtained by searching in the constructed latent space, the attribute combination of the detection target image is taken as an initial population, and the genetic algorithm is used to recombine the attribute combination to obtain a recombined image attribute combination;

[0020] The reconstructed image is obtained by image reconstruction according to the recombined attribute combination, and a plurality of SDAE layers are stacked to learn and extract learning features of the reconstructed image layer by layer to generate a reconstructed feature map;

[0021] The reconstructed feature map is taken as an input of an underwater target recognition model to identify and analyze the detected target, and underwater target recognition information is obtained.

[0022] In the scheme, the underwater target recognition information is acquired, and target tracking is performed according to the underwater target recognition information. The motion path of the yellowfin tuna and the fertilized egg in a unit time is analyzed to obtain motion path analysis information, which specifically includes:

[0023] The underwater target recognition information is acquired, and target frame images are extracted according to the underwater target recognition information. The target frame images are fertilized egg target frame images and yellowfin tuna target frame images.

[0024] According to the target frame images, a target tracking algorithm is used to track the motion trajectory of the identified target. Based on the time attribute of the target frame images, a monitoring image in the next time is obtained, and a next detection target image is obtained through target detection.

[0025] The cosine similarity between the next detection target image and the target frame image is calculated, and whether it is a tracking target is determined according to the cosine similarity. If it is, it is marked as a tracking target.

[0026] A three-dimensional coordinate system is constructed based on the size of the target aquaculture net cage and the position of the underwater monitoring facility, and the initial spatial coordinates and the tracking spatial coordinates of the tracking target are obtained.

[0027] An extended Kalman filter is used to estimate the motion state of the tracking target, and a motion prediction box is generated according to the running state estimation result. The matching degree between the target frame image and the image in the motion prediction box is calculated.

[0028] Based on the calculated matching degree, it is determined whether the predicted target is a monitoring target. If it is a monitoring target, the spatial coordinate information of the predicted target is obtained through the constructed three-dimensional coordinate system, the motion path is predicted in combination with the initial spatial coordinates, the predicted path is corrected using the tracking spatial coordinates, and the motion path analysis information is obtained.

[0029] In the scheme, the water flow field simulation model is constructed, the water flow velocity monitoring information is acquired, the water flow field condition in the target aquaculture net cage is analyzed, and the fertilized egg motion trajectory prediction is performed to obtain fertilized egg motion trajectory prediction information, which specifically includes:

[0030] The target aquaculture net cage structure information is acquired, and the water flow field simulation model is constructed according to the target aquaculture net cage structure information. The water flow field simulation model is subjected to grid processing.

[0031] The water body flow velocity monitoring technology is used to monitor the water body flow velocity of the target aquaculture net cage to obtain water body flow velocity monitoring information, boundary conditions are set, the water body flow velocity monitoring information is taken as input of a water body flow field simulation model, a fluid solver is used for solving, and water body flow field simulation information is obtained;

[0032] The flow velocity data is converted into a flow velocity vector field according to the water body flow field simulation information, and a three-dimensional vector diagram is generated, and flow velocity vector features of different regions are obtained based on the three-dimensional vector diagram;

[0033] Motion path analysis information is obtained, the motion path trend of the yellowfin tuna and the fertilized eggs in the target aquaculture net is analyzed, and motion path trend information is obtained; the motion state is estimated based on the motion path trend information, the motion speed and the motion direction of the yellowfin tuna and the fertilized eggs are analyzed, and first analysis information is obtained;

[0034] The fertilized egg motion trajectory prediction information is obtained by combining the flow velocity vector features of different regions and the first analysis information with a particle tracking algorithm, defining the initial position of the target fertilized egg, setting initial conditions based on the flow velocity vector features and the first analysis information, and performing iterative analysis.

[0035] In the scheme, whether the target fertilized egg is at risk is judged based on the fertilized egg motion trajectory prediction information, risk warning information is generated, and a control scheme is developed, specifically including:

[0036] The fertilized egg motion trajectory prediction information is obtained, the motion position features and the time features of the target fertilized egg in the future time period are extracted according to the fertilized egg motion trajectory prediction information, and first feature information is obtained;

[0037] The first analysis information is obtained, the motion state features of the yellowfin tuna in the target aquaculture net are extracted based on the first analysis information, the motion route of the yellowfin tuna is generated, and collision risk analysis is performed in combination with the first feature information;

[0038] The boundary position information of the target aquaculture net cage is obtained, the first feature information is operated with the boundary position, the relative position distance between the fertilized egg and the net cage boundary in each time period is obtained, and first relative position distance information is obtained;

[0039] The first risk threshold is set, the first relative position distance information is judged with the first risk threshold, whether there is a collision risk with the net cage boundary is analyzed, and first judgment result information is obtained;

[0040] The motion path position features and the time features of the yellowfin tuna are extracted based on the motion route of the yellowfin tuna, and the first feature information is compared to judge whether there is a path intersection condition in the same time;

[0041] If there is a path intersection condition, analyze whether there is a collision condition between the yellowfin tuna and the fertilized egg, extract the collision position characteristics and calculate the relative position distance of the yellowfin tuna and the fertilized egg at the intersection position, to obtain second relative position distance information;

[0042] A second risk threshold is set, and the second relative position distance information is judged with the second risk threshold to analyze whether there is a collision, to obtain second judgment result information;

[0043] A preset control strategy database is set, and the fertilized egg risk warning information is generated based on the first judgment result information and the second judgment result information, and the control strategy is obtained from the preset control strategy database for prevention and control.

[0044] The second aspect of the application provides a prevention and control system for tuna fertilized egg overflow and biological predation, which comprises a memory and a processor, wherein the memory contains a prevention and control method program based on tuna fertilized egg overflow and biological predation, and the prevention and control method program based on tuna fertilized egg overflow and biological predation is executed by the processor to realize the following steps:

[0045] Obtain underwater real-time monitoring information of the target culture net cage, perform target detection based on the underwater real-time monitoring information, and obtain target detection information;

[0046] Construct an underwater target recognition model, recognize the detected target based on the target detection information, recognize the yellowfin tuna and the fertilized egg, and obtain underwater target recognition information;

[0047] Obtain the underwater target recognition information, track the target according to the underwater target recognition information, analyze the movement path of the yellowfin tuna and the fertilized egg in unit time, and obtain movement path analysis information;

[0048] Construct a water flow field simulation model, obtain water flow velocity monitoring information, analyze the water flow field condition in the target culture net cage, and perform fertilized egg movement trajectory prediction, to obtain fertilized egg movement trajectory prediction information;

[0049] Judge whether the target fertilized egg is at risk based on the fertilized egg movement trajectory prediction information, generate risk warning information and develop a control scheme.

[0050] The application discloses a method and system for preventing and controlling tuna fertilized egg overflow and biological predation, and comprises the following steps: obtaining underwater real-time monitoring information, performing target detection based on the underwater real-time monitoring information to obtain target detection information; identifying the detection target based on the target detection information to identify yellowfin tuna and fertilized eggs and obtain underwater target identification information; performing target tracking according to the underwater target identification information, analyzing the movement path of the yellowfin tuna and the fertilized eggs in a unit time to obtain movement path analysis information; constructing a water flow field simulation model, analyzing the water flow field condition in the target culture net cage, and performing fertilized egg movement trajectory prediction to obtain fertilized egg movement trajectory prediction information; judging whether the target fertilized egg is at risk based on the fertilized egg movement trajectory prediction information, generating risk warning information and formulating a control scheme. The yield and quality of tuna culture are effectively improved. BRIEF DESCRIPTION OF DRAWINGS

[0051] In order to more clearly illustrate the technical solutions in the embodiments or examples of the present application, the following will briefly introduce the drawings needed to be used in the embodiments or examples. Obviously, the drawings in the following description only some embodiments of the present application, and for those skilled in the art, other drawings can also be obtained without creative labor.

[0052] Figure 1 A flow chart of a method for preventing and controlling tuna fertilized egg overflow and biological predation is provided for an embodiment of the present application.

[0053] Figure 2 A flow chart of a method for preventing and controlling tuna fertilized egg overflow and biological predation is provided for an embodiment of the present application.

[0054] Figure 3 A flow chart of a method for preventing and controlling tuna fertilized egg overflow and biological predation is provided for an embodiment of the present application.

[0055] The implementation, functional features and advantages of the present application will be further described with reference to the drawings. DETAILED DESCRIPTION

[0056] In order to more clearly illustrate the technical solutions in the embodiments or examples of the present application, the following will briefly introduce the drawings needed to be used in the embodiments or examples. Obviously, the drawings in the following description only some embodiments of the present application, and for those skilled in the art, other drawings can also be obtained without creative labor.

[0057] In the following description, many specific details are set forth in order to provide a thorough understanding of the present application, however, the present application can also be implemented in other ways different from those described herein, therefore, the protection scope of the present application is not limited by the specific embodiments disclosed below.

[0058] Figure 1 A flow chart of a prevention and control method of tuna fertilized egg overflow and biological predation is provided for an embodiment of the present application;

[0059] As shown in Figure 1 The present application provides a flow chart of a prevention and control method of tuna fertilized egg overflow and biological predation, comprising:

[0060] S102, obtaining underwater real-time monitoring information of a target culture net cage, performing target detection based on the underwater real-time monitoring information, and obtaining target detection information;

[0061] S104, constructing an underwater target identification model, identifying the detected target based on the target detection information, identifying yellowfin tuna and fertilized eggs, and obtaining underwater target identification information;

[0062] S106, obtaining underwater target identification information, performing target tracking according to the underwater target identification information, analyzing the movement path of yellowfin tuna and fertilized eggs within a unit time, and obtaining movement path analysis information;

[0063] S108, constructing a water flow field simulation model, obtaining water flow velocity monitoring information, analyzing the water flow field condition in the target culture net cage, and performing fertilized egg movement trajectory prediction, and obtaining fertilized egg movement trajectory prediction information;

[0064] S110, judging whether the target fertilized egg is at risk based on the fertilized egg movement trajectory prediction information, generating risk warning information and formulating a control scheme.

[0065] It should be noted that the present application provides a prevention and control method and system for tuna fertilized egg overflow and biological predation, target detection is performed based on underwater real-time monitoring information, useful images in the monitoring images are screened out, target detection information is obtained; the detected target is identified based on the target detection information, yellowfin tuna and fertilized eggs are identified, and further useless detection targets are removed, underwater target identification information is obtained; target tracking is performed according to the underwater target identification information, the movement path of yellowfin tuna and fertilized eggs within a unit time is analyzed, the movement route of fertilized eggs and yellowfin tuna in actual conditions is analyzed, movement path analysis information is obtained; a water flow field simulation model is constructed, the water flow field condition in the target culture net cage is analyzed, and fertilized egg movement trajectory prediction is performed, the water flow velocity factor and the swimming characteristics of yellowfin tuna are considered for analysis, the fertilized egg movement trajectory is more consistent with the actual condition, fertilized egg movement trajectory prediction information is obtained; whether the target fertilized egg is at risk is judged based on the fertilized egg movement trajectory prediction information, risk warning information is generated and a control scheme is formulated, thereby improving the yield and quality of tuna culture.

[0066] Further, in a preferred embodiment of the present application, the underwater real-time monitoring information of the target aquaculture net cage is obtained, target detection is performed based on the underwater real-time monitoring information, and target detection information is obtained, specifically including:

[0067] The target aquaculture net cage is monitored, underwater real-time monitoring information of the target aquaculture net cage is obtained based on underwater video monitoring technology and image preprocessing is performed, and preprocessed image information is obtained;

[0068] A target detection model is constructed based on YOLOv5, the preprocessed image information is input for target detection, feature extraction is performed on the preprocessed image information, and an initial feature map is generated;

[0069] Scale features corresponding to each feature are obtained according to the initial feature map, attention scores of each feature are obtained by introducing an attention mechanism, an attention feature map is generated, and the initial feature map and the attention feature map are fused by using a feature fusion pyramid;

[0070] A candidate detection box is generated according to the fused feature map, non-maximum suppression is used for screening, and a final target detection box is obtained, and target detection information is obtained through the final target detection box.

[0071] It should be noted that the aquaculture net cage is monitored by underwater video monitoring technology, the cultivation status of the yellowfin tuna in the net cage is obtained in real time, and then the images of the objects existing in the underwater real-time monitoring information are identified through target detection, and useful information is screened out.

[0072] Further, in a preferred embodiment of the present application, the underwater target recognition model is constructed, the detection target is recognized based on the target detection information, the yellowfin tuna and the fertilized egg are recognized, and underwater target recognition information is obtained, specifically including:

[0073] The fish body image and the fertilized egg image of the underwater yellowfin tuna are obtained based on data retrieval, the obtained image information is preprocessed, sample expansion processing is performed according to the preprocessed image, and an instance data set is formed;

[0074] The underwater target recognition model is constructed through a neural network, and the underwater target recognition model is deep learning and trained by using the instance data set to construct a training data set;

[0075] The target detection information is obtained, the detection target image is obtained by using an image segmentation algorithm, and is input into a stacked denoising autoencoder for encoding, a low-dimensional latent representation of the detection target image is obtained and a latent space is constructed;

[0076] Search in the constructed latent space to obtain attribute combinations of the detection target image, take the attribute combinations of the detection target image as an initial population, and use a genetic algorithm to recombine the attribute combinations to obtain recombined image attribute combinations;

[0077] Reconstruct an image according to the recombined attribute combinations to obtain a reconstructed image, perform layer-by-layer learning on the reconstructed image by stacking multiple SDAE layers, and extract learning features to generate a reconstructed feature map;

[0078] Take the reconstructed feature map as an input of an underwater target recognition model, recognize and analyze the detection target to obtain underwater target recognition information.

[0079] It should be noted that the fish body image and the image of the fertilized egg of the yellowfin tuna are obtained through data retrieval and preprocessed, including denoising, normalization, contrast enhancement, etc., to improve the image quality and remove redundant information. More samples are generated through data augmentation technology to form an instance dataset containing different angles and different lighting conditions. An underwater target recognition model is built using a neural network, which can be a residual neural network. Based on the target detection information, an image segmentation algorithm is used to extract the image region of the detection target, and the segmented image is input into a stacked denoising autoencoder (SDAE) for encoding processing. Through the encoding process, the image of the detection target is mapped to a low-dimensional latent representation, and a latent space is constructed therefrom. In the latent space, a search mechanism, which can be an average search algorithm, is used to obtain attribute combinations of the detection target image, such as shape, color, texture, etc. Then, a genetic algorithm is used to recombine these attribute combinations. By simulating genetic selection and mutation processes, the genetic algorithm can generate new attribute combinations, ensuring that the features of the image are more diverse and producing different recombined image attribute combinations. Based on these combinations, the image is reconstructed to generate a new reconstructed image. Multiple SDAE layers are stacked to perform layer-by-layer learning on the reconstructed image and extract learning features to generate a reconstructed feature map representing the features of the target object. Finally, the generated reconstructed feature map is taken as an input of the underwater target recognition model for recognition and analysis of the target object. Thus, the required images of the yellowfin tuna and the fertilized egg are obtained.

[0080] Further, in a preferred embodiment of the present application, the underwater target recognition information is obtained, target tracking is performed according to the underwater target recognition information, the motion path of the yellowfin tuna and the fertilized egg in a unit time is analyzed, and motion path analysis information is obtained, specifically including:

[0081] The underwater target recognition information is obtained, and target frame images are extracted according to the underwater target recognition information, the target frame images being fertilized egg target frame images and yellowfin tuna target frame images;

[0082] According to the target frame image, a target tracking algorithm is used to track the motion trajectory of the identified target, a monitoring image in the next time is obtained based on the time attribute of the target frame image, and a next detection target image is obtained through target detection;

[0083] A cosine similarity between the next detection target image and the target frame image is calculated, and it is judged whether the target is a tracking target according to the cosine similarity, and if so, the target is marked as a tracking target;

[0084] A three-dimensional coordinate system is constructed based on the size of the target aquaculture net cage and the position of the underwater monitoring facility, and initial spatial coordinates and tracking spatial coordinates of the tracking target are obtained;

[0085] An extended Kalman filter is used to estimate the motion state of the tracking target, a motion prediction box is generated according to the estimation result, and the matching degree between the target frame image and the image in the motion prediction box is calculated;

[0086] It is judged whether the predicted target is a monitoring target based on the calculated matching degree, if it is a monitoring target, the spatial coordinate information of the predicted target is obtained through the constructed three-dimensional coordinate system, the motion path is predicted in combination with the initial spatial coordinates, the predicted path is corrected using the tracking spatial coordinates, and the motion path analysis information is obtained.

[0087] It should be noted that after the underwater target is identified, the target frame image of the yellowfin tuna and the fertilized egg is extracted, and the target frame image is a key frame image of the identified yellowfin tuna or fertilized egg. The target tracking algorithm is used to track the motion trajectory of the identified target. According to the time attribute of the target frame image, the image at the next time point is obtained from the continuous monitoring image, new target detection is performed, and the subsequent detection image of the detection target is obtained. The cosine similarity between the next detection target image and the target frame image is calculated to determine whether the two images belong to the same tracking target. The cosine similarity is a measure for measuring the similarity between the feature vectors of two images. If the similarity is high, the target can be determined as the target being tracked, and it is marked as a tracking target. Then, a three-dimensional coordinate system is constructed, and the initial spatial coordinates of the tracking target and the tracking spatial coordinates changing over time are obtained. The extended Kalman filter is used to estimate the motion state of the tracking target. The Kalman filter can predict the motion of the target in a dynamic environment and generate a motion prediction box. The system extracts the image in the motion prediction box and calculates the matching degree of the image and the original target frame image. According to the calculated matching degree, it is judged whether the target in the prediction box is the monitoring target. If the matching degree is high, it can be considered that the prediction target is the actual monitoring target. At this time, the spatial coordinate information of the prediction target is obtained by using the constructed three-dimensional coordinate system, and the motion path prediction is performed in combination with the initial spatial coordinates of the target. Finally, the predicted path is corrected by the tracking spatial coordinates, so that more accurate motion path analysis information is obtained. Thus, the motion trajectory of the continuously monitored target in the underwater environment is understood, and data basis is provided for subsequent motion trajectory prediction.

[0088] Further, in a preferred embodiment of the present application, the water flow field simulation model is constructed, the water flow velocity monitoring information is obtained, the water flow field condition in the target aquaculture net cage is analyzed, and the fertilized egg motion trajectory prediction is performed to obtain the fertilized egg motion trajectory prediction information, specifically comprising:

[0089] Obtain the target aquaculture net cage structure information, construct the water flow field simulation model according to the target aquaculture net cage structure information, and perform grid processing on the water flow field simulation model;

[0090] Based on the water flow velocity monitoring technology, the water flow velocity of the target aquaculture net cage is monitored to obtain the water flow velocity monitoring information, the boundary conditions are set, the water flow velocity monitoring information is taken as the input of the water flow field simulation model, the fluid solver is used for solving, and the water flow field simulation information is obtained;

[0091] According to the water flow field simulation information, the flow velocity data is converted into a flow velocity vector field and a three-dimensional vector diagram is generated, and the flow velocity vector features of different regions are obtained based on the three-dimensional vector diagram;

[0092] Obtaining motion path analysis information, analyzing the motion path trend of the yellowfin tuna and the fertilized eggs inside the target culture net, and obtaining motion path trend information; estimating the motion state based on the motion path trend information, analyzing the motion speed and the motion direction of the yellowfin tuna and the fertilized eggs, and obtaining first analysis information;

[0093] Combining the flow velocity vector characteristics of different regions and the first analysis information, using a particle tracking algorithm to predict the motion trajectory of the fertilized eggs, defining the initial position of the target fertilized eggs, setting initial conditions based on the flow velocity vector characteristics and the first analysis information, and performing iterative analysis to obtain fertilized egg motion trajectory prediction information.

[0094] It should be noted that a water flow field simulation model is established based on the structure information of the target culture net cage. The flow field model is gridded according to the actual structure of the net cage, so that the simulation model can accurately reflect the flow of water in the culture net cage. Next, the water in the culture net cage is monitored based on water flow velocity monitoring technology, and water flow velocity monitoring data is obtained. The boundary conditions are set, the flow velocity monitoring information is input into the water flow field simulation model, and the model is solved by a fluid solver, such as ANSYS Fluent, to obtain the simulation results of the water flow field. The flow velocity data is converted into a flow velocity vector field, and a three-dimensional vector diagram is generated. Thus, the changes of water flow velocity and flow direction in different regions of the culture net cage are intuitively reflected. Based on the three-dimensional vector diagram, the flow velocity vector characteristics of each region are analyzed to determine the flow characteristics of different water areas in the net cage, including important information such as vortex, dead water area and main flow direction. Subsequently, the motion trend of the yellowfin tuna and the fertilized eggs in the culture net cage is analyzed based on the motion path analysis information, which can be analyzed by a trend analysis model such as ARIMA model to obtain motion path trend information, and the motion state is estimated based on this, including motion speed and motion direction. The actual motion behavior of the fertilized eggs and the tuna in the water body under the influence of flow velocity and flow direction is understood. Subsequently, the motion trajectory of the fertilized eggs is predicted using a particle tracking algorithm in combination with the characteristics of the water flow velocity vector field and the first analysis information. The initial position of the fertilized eggs is defined, and the initial conditions are set based on the flow velocity vector characteristics of the water body and the previously analyzed motion state information. The motion trajectory of the fertilized eggs in the water body is updated iteratively, and finally the fertilized egg motion trajectory prediction information is obtained. Considering the water flow velocity and the swimming characteristics of the yellowfin tuna, the predicted trajectory is more consistent with the actual situation.

[0095] Further, in a preferred embodiment of the present application, the method further comprises:

[0096] Obtaining fertilized egg motion trajectory prediction information, extracting the motion position feature and the time feature of the target fertilized egg reaching the corresponding position in the future time period according to the fertilized egg motion trajectory prediction information, and obtaining first feature information;

[0097] Obtaining first analysis information, extracting the motion state feature of the yellowfin tuna inside the target culture net based on the first analysis information, generating the motion route of the yellowfin tuna, and performing collision risk analysis in combination with the first feature information;

[0098] Obtaining the boundary position information of the target culture net cage, operating the first feature information and the boundary position, obtaining the relative position distance of the fertilized egg and the net cage boundary in each time period, and obtaining first relative position distance information;

[0099] Setting a first risk threshold, judging the first relative position distance information and the first risk threshold, analyzing whether there is a collision risk with the net cage boundary, and obtaining first judgment result information;

[0100] Extracting the motion path position feature and time feature of the yellowfin tuna based on the motion route of the yellowfin tuna, comparing the first feature information, and judging whether there is a path intersection condition in the same time;

[0101] If there is a path intersection condition, analyzing whether there is a collision condition between the yellowfin tuna and the fertilized egg, extracting the collision position feature and calculating the relative position distance of the yellowfin tuna and the fertilized egg at the intersection position, and obtaining second relative position distance information;

[0102] Setting a second risk threshold, judging the second relative position distance information and the second risk threshold, analyzing whether there is a collision, and obtaining second judgment result information;

[0103] Presetting a control strategy database, generating fertilized egg risk warning information based on the first judgment result information and the second judgment result information, and obtaining a control strategy from the preset control strategy database for prevention and control.

[0104] It should be noted that the movement position characteristics of the fertilized egg in the future time period and the time characteristics of the fertilized egg reaching these positions are extracted through the fertilized egg movement trajectory prediction information to form first feature information. Subsequently, the movement state characteristics of the yellowfin tuna are extracted based on the first analysis information, and the movement route of the yellowfin tuna is generated. In combination with the first feature information, the collision risk that may occur to the fertilized egg is evaluated. The boundary position information of the target net cage is obtained, the movement position characteristics of the fertilized egg are operated with the boundary position of the net cage, the relative distance between the fertilized egg and the boundary of the net cage in each time period is determined, and first relative position distance information is generated. A first risk threshold is set and compared with the first risk threshold to determine whether the fertilized egg has a risk of collision with the boundary of the net cage, and first judgment result information is obtained. Subsequently, the spatial position and time characteristics of the movement path of the yellowfin tuna are extracted through the movement route of the yellowfin tuna, and are compared with the first feature information of the fertilized egg to determine whether the movement paths of the two at the same time intersect. If there is a path intersection, it is further analyzed whether a collision between the yellowfin tuna and the fertilized egg is possible, the possible collision position characteristics are extracted, and the relative distance between the yellowfin tuna and the fertilized egg at the intersection position is calculated, thereby generating second relative position distance information. A second risk threshold is set, and the second relative position distance information is compared with the threshold to obtain second judgment result information. Finally, the risk warning information of the fertilized egg is generated according to the first judgment result information and the second judgment result information, and a suitable control strategy is selected from the control strategy database to prevent the fertilized egg from overflowing and the collision between the yellowfin tuna and the fertilized egg, to ensure the safety of the fertilized egg and the fish population during the cultivation process, to reduce the risk of the fertilized egg overflowing and being preyed upon, and to improve the success rate and efficiency of the cultivation.

[0105] Figure 2 A fertilized egg monitoring process chart of a net cage is provided for an embodiment of the present application;

[0106] As Figure 2 shown, the present application provides a fertilized egg monitoring process chart of a net cage, which comprises:

[0107] S202, underwater real-time monitoring information is obtained based on a set underwater video monitoring array, target detection and underwater target identification are performed according to the underwater real-time monitoring information, and underwater target identification information is obtained;

[0108] S204, the fertilized egg target frame image and the yellowfin tuna target frame image are extracted according to the underwater identification target information, target motion tracking is performed, the movement path of the yellowfin tuna and the fertilized egg in a unit time is analyzed, and movement path analysis information is obtained;

[0109] S206, the water flow field condition in the target net cage is analyzed based on a water flow speed monitoring technology, and the fertilized egg movement trajectory prediction is performed in combination with the movement path analysis information, and fertilized egg movement trajectory prediction information is obtained;

[0110] S208, judging whether the target fertilized egg has a risk based on the fertilized egg motion trajectory prediction information, generating risk warning information and formulating a regulation scheme.

[0111] It should be noted that the underwater video monitoring array is set to obtain underwater monitoring information in the net cage in real time, and target detection and identification are performed based on the monitoring information. By analyzing the video data, the underwater target objects, including the yellowfin tuna and the fertilized eggs, are identified, and underwater target identification information is generated to eliminate useless detection images. Next, using the underwater target identification information, the target frame images of the fertilized eggs and the yellowfin tuna are extracted for detailed target motion tracking analysis to understand their actual motion law, providing initial conditions for subsequent fertilized egg motion trajectory prediction, including motion speed and direction. Subsequently, combined with the water flow rate monitoring technology, the water flow field condition inside the net cage is analyzed to evaluate the speed and direction of the water flow. Then, through the motion path analysis information, the motion trajectory of the fertilized egg is predicted, considering the flow characteristics of the water body inside the net cage and the normal swimming characteristics of the yellowfin tuna, to further improve the prediction accuracy and precision. Finally, based on the motion trajectory prediction information of the fertilized egg, it is judged whether the target fertilized egg has any risk of collision or overflow. Once a possible risk is detected, risk warning information is generated, and a corresponding regulation scheme is formulated based on the warning information to effectively control the risk and ensure the safety of the fertilized eggs during the breeding process, reducing the risk of fertilized egg overflow and predation, and improving the success rate and efficiency of breeding.

[0112] Figure 3 A tuna fertilized egg overflow and biological predation prevention and control system 3 is provided for an embodiment of the present application, which comprises a memory 31 and a processor 32. The memory 31 contains a tuna fertilized egg overflow and biological predation prevention and control method program, which is executed by the processor 32 to implement the following steps:

[0113] Obtain underwater real-time monitoring information of the target net cage, and perform target detection based on the underwater real-time monitoring information to obtain target detection information.

[0114] Construct an underwater target identification model, identify the detection target based on the target detection information, identify the yellowfin tuna and the fertilized eggs, and obtain underwater target identification information.

[0115] Obtain underwater target identification information, track the target according to the underwater target identification information, analyze the motion path of the yellowfin tuna and the fertilized eggs in a unit of time, and obtain motion path analysis information.

[0116] A water body flow field simulation model is constructed to obtain water flow monitoring information, analyze the water flow field condition in the target net cage, and predict the fertilized egg movement trajectory to obtain fertilized egg movement trajectory prediction information.

[0117] Based on the fertilized egg movement trajectory prediction information, it is determined whether the target fertilized egg is at risk, risk warning information is generated, and a control scheme is developed.

[0118] In several embodiments provided in the present application, it should be understood that the disclosed devices and methods can be implemented in other ways. The device embodiments described above are only illustrative, for example, the division of the units is only a logical function division, and actual implementation can have another division manner, such as: multiple units or components can be combined, or can be integrated into another system, or some features can be ignored or not executed. In addition, the coupling or direct coupling or communication connection between the various components shown or discussed can be through some interface, indirect coupling or communication connection of devices or units, which can be electrical, mechanical or other forms.

[0119] The units described above as separate components can or can not be physically separated, and the components shown as units can or can not be physical units; they can be located in one place or distributed on multiple network units; part or all of the units can be selected according to actual needs to achieve the purpose of the embodiment scheme.

[0120] In addition, each functional unit in each embodiment of the present application can be integrated into one processing unit, or each unit can be a separate unit, or two or more units can be integrated into one unit; the integrated unit can be realized in the form of hardware or hardware plus software functional unit.

[0121] Those of ordinary skill in the art can understand that all or part of the steps of the above method embodiments can be completed by program instruction related hardware, and the foregoing program can be stored in a computer readable storage medium, and the program executes the steps including the above method embodiments when executed; and the foregoing storage medium includes: mobile storage device, read-only memory (ROM, Read-Only Memory), random access memory (RAM, Random Access Memory), magnetic disk or optical disk, and various storage program codes.

[0122] Alternatively, the above-mentioned integrated unit of the present application, if realized in the form of a software function module and sold or used as an independent product, can also be stored in a computer-readable storage medium. Based on such an understanding, the technical solutions of the embodiments of the present application can be embodied in the form of a software product, and the computer software product is stored in a storage medium, including a number of instructions to make a computer device (which can be a personal computer, a server, or a network device, etc.) execute all or part of the methods described in the embodiments of the present application. The aforementioned storage medium includes: mobile storage devices, ROM, RAM, magnetic disks or optical disks, and various media that can store program codes.

[0123] The above is only a specific implementation of the present application, but the protection scope of the present application is not limited thereto, any person skilled in the art can easily think of changes or replacements within the technical range disclosed by the present application, which should be covered within the protection scope of the present application. Therefore, the protection scope of the present application should be subject to the protection scope of the claims.

Claims

1. A method for the prevention and control of the external spillage and biological predation of tuna fertilized eggs, characterized in that, The method comprises the following steps: obtaining underwater real-time monitoring information of a target culture cage, performing target detection on underwater real-time detection image information through a YOLOv5 target detection model, and obtaining a target detection frame as target detection information; constructing an underwater target recognition model, identifying the detection target based on the target detection information, identifying yellowfin tuna and fertilized eggs, and obtaining underwater target recognition information; obtaining underwater target recognition information, tracking the target according to the underwater target recognition information, analyzing the movement path of the yellowfin tuna and the fertilized eggs in a unit time, and obtaining movement path analysis information; constructing a water flow field simulation model according to the structure of the target culture cage, obtaining water flow velocity monitoring information and inputting the water flow velocity monitoring information into the water flow field simulation model, analyzing the flow velocity vector characteristics of different regions, combining the path analysis information to estimate the movement state of the yellowfin tuna and the fertilized eggs and predict the movement trajectory of the fertilized eggs, and obtaining fertilized egg movement trajectory prediction information; judging whether the target fertilized egg has a collision risk based on the fertilized egg movement trajectory prediction information, generating risk warning information, and formulating a control scheme.

2. A method of preventing and controlling the external spillage and biological predation of tuna fertilized eggs according to claim 1, characterized in that, The method comprises the following steps: monitoring the target culture cage, obtaining underwater real-time monitoring information of the target culture cage based on underwater video monitoring technology, and performing image preprocessing to obtain preprocessed image information; constructing a target detection model based on YOLOv5, inputting the preprocessed image information for target detection, performing feature extraction on the preprocessed image information, and generating an initial feature map; obtaining scale features corresponding to each feature according to the initial feature map, introducing an attention mechanism to obtain attention scores of each feature, generating an attention feature map, and fusing the initial feature map and the attention feature map through a feature fusion pyramid; generating a candidate detection frame according to the fused feature map, screening through a non-maximum suppression method, obtaining a final target detection frame, and obtaining target detection information through the final target detection frame.

3. A method of preventing and controlling the external spillage and biological predation of tuna fertilized eggs according to claim 1, characterized in that, The method comprises the following steps: obtaining fish body images and fertilized egg images of underwater yellowfin tuna based on data retrieval, preprocessing the obtained image information, performing sample expansion processing according to the preprocessed images, and constituting an instance data set; constructing an underwater target recognition model through a neural network, constructing a training data set using the instance data set, and performing deep learning and training on the underwater target recognition model; obtaining target detection information, obtaining a detection target image using an image segmentation algorithm, inputting the detection target image into a stacked denoising autoencoder for encoding, obtaining a low-dimensional latent representation of the detection target image and constructing a latent space; searching in the constructed latent space to obtain attribute combinations of the detection target image, taking the attribute combinations of the detection target image as an initial population, and using a genetic algorithm to recombine the attribute combinations to obtain recombined attribute combinations; The reconstructed image is obtained by image reconstruction according to the combination of the reconstruction attributes, and the reconstructed image is learned layer by layer and the learning features are extracted by stacking multiple SDAE layers to generate a reconstructed feature map; The reconstructed feature map is taken as the input of the underwater target recognition model to recognize and analyze the detected target, and underwater target recognition information is obtained.

4. The method of claim 1, wherein the method is characterized by, The underwater target recognition information is obtained, and target frame images are extracted according to the underwater target recognition information, the target frame images being fertilized egg target frame images and yellowfin tuna target frame images; According to the target frame images, a target tracking algorithm is used to track the motion trajectory of the recognized target, a monitoring image in the next time is obtained based on the time attribute of the target frame images, and a next detection target image is obtained through target detection; The cosine similarity between the next detection target image and the target frame image is calculated, and it is judged whether it is a tracking target according to the cosine similarity; if yes, it is marked as a tracking target; A three-dimensional coordinate system is constructed based on the size of the target aquaculture net cage and the position of the underwater monitoring facility, and the initial spatial coordinates and the tracking spatial coordinates of the tracking target are obtained; An extended Kalman filter is used to estimate the motion state of the tracking target, a motion prediction box is generated according to the running state estimation result, and the matching degree between the image in the motion prediction box and the target frame image is calculated; It is judged whether the predicted target is a monitoring target based on the calculated matching degree; if it is a monitoring target, the spatial coordinate information of the predicted target is obtained through the constructed three-dimensional coordinate system, the motion path is predicted in combination with the initial spatial coordinates, the tracking spatial coordinates are used to correct the predicted path, and the motion path analysis information is obtained. The water flow field simulation model is constructed according to the structure of the target aquaculture net cage, the water flow velocity monitoring information is obtained and input into the water flow field simulation model, the flow velocity vector characteristics of different regions are analyzed, the motion state of the yellowfin tuna and the fertilized eggs is estimated and the motion trajectory of the fertilized eggs is predicted in combination with the path analysis information, and the fertilized egg motion trajectory prediction information is obtained, specifically including:

5. The method of claim 1, wherein the method is characterized by: The target aquaculture net cage structure information is obtained, and the water flow field simulation model is constructed according to the target aquaculture net cage structure information, and the water flow field simulation model is processed by gridding; The water flow velocity monitoring information is obtained by monitoring the water flow velocity of the target aquaculture net cage based on the water flow velocity monitoring technology, the boundary conditions are set, the water flow velocity monitoring information is taken as the input of the water flow field simulation model, the fluid solver is used for solving, and the water flow field simulation information is obtained; According to the water flow field simulation information, the flow velocity data is converted into a flow velocity vector field and a three-dimensional vector diagram is generated, and the flow velocity vector characteristics of different regions are obtained based on the three-dimensional vector diagram; ​ Obtain motion path analysis information, analyze the motion path trend of the yellowfin tuna and the fertilized eggs inside the target culture net, and obtain motion path trend information; estimate the motion state based on the motion path trend information, analyze the motion speed and direction of the yellowfin tuna and the fertilized eggs, and obtain first analysis information; Combine the flow velocity vector characteristics of different regions and the first analysis information to use a particle tracking algorithm to predict the motion trajectory of the fertilized eggs, define the initial position of the target fertilized eggs, set the initial conditions based on the flow velocity vector characteristics and the first analysis information, and perform iterative analysis to obtain fertilized egg motion trajectory prediction information.

6. A method of preventing and controlling the external spillage and biological predation of tuna fertilized eggs according to claim 1, characterized in that, The method comprises the following steps: Obtain the motion trajectory prediction information of the fertilized eggs, extract the motion position characteristics and the time characteristics of the target fertilized eggs reaching the corresponding position in the future time period according to the motion trajectory prediction information of the fertilized eggs, and obtain first feature information; Obtain the first analysis information, extract the motion state characteristics of the yellowfin tuna inside the target culture net based on the first analysis information, generate the motion route of the yellowfin tuna, and perform collision risk analysis in combination with the first feature information; Obtain the boundary position information of the target culture net cage, operate the first feature information and the boundary position, obtain the relative position distance between the fertilized eggs and the net cage boundary in each time period, and obtain first relative position distance information; Set a first risk threshold, judge the first relative position distance information and the first risk threshold, analyze whether there is a collision risk with the net cage boundary, and obtain first judgment result information; Compare the motion path position characteristics and time characteristics of the yellowfin tuna with the first feature information based on the motion route of the yellowfin tuna, and judge whether there is a path intersection condition in the same time; If there is a path intersection condition, analyze whether there is a collision condition between the yellowfin tuna and the fertilized eggs, extract the collision position characteristics and calculate the relative position distance between the yellowfin tuna and the fertilized eggs at the intersection position, and obtain second relative position distance information; Set a second risk threshold, judge the second relative position distance information and the second risk threshold, analyze whether there is a collision, and obtain second judgment result information; Pre-set a control strategy database, generate fertilized egg risk warning information based on the first judgment result information and the second judgment result information, and obtain a control strategy from the pre-set control strategy database for prevention and control.

7. A system for the prevention and control of external spillage and biological predation of tuna fertilized eggs, characterized by, The system comprises a memory and a processor, the memory contains tuna sperm egg overflow and biological predation prevention and control method program instructions, and the tuna sperm egg overflow and biological predation prevention and control method program instructions are executed by the processor to realize the following steps: Obtain underwater real-time monitoring information of the target culture net cage, perform target detection on the underwater real-time detection image information through a YOLOv5 target detection model, and obtain a target detection box as target detection information; The underwater target recognition model is constructed, the detection target is recognized based on the target detection information, the bigeye tuna and the fertilized egg are recognized, and underwater target recognition information is obtained; The underwater target recognition information is obtained, target tracking is performed according to the underwater target recognition information, the motion path of the bigeye tuna and the fertilized egg in a unit time is analyzed, and motion path analysis information is obtained; According to the structure of the target culture net cage, a water flow field simulation model is constructed, water flow velocity monitoring information is obtained and input into the water flow field simulation model, the flow velocity vector characteristics of different regions are analyzed, the motion state of the bigeye tuna and the fertilized egg is estimated and the motion trajectory of the fertilized egg is predicted in combination with the path analysis information, and fertilized egg motion trajectory prediction information is obtained; Based on the fertilized egg motion trajectory prediction information, it is judged whether the target fertilized egg has a collision risk, risk warning information is generated, and a control scheme is developed.

Images