A method for evaluating the aquaculture capacity of a submersible and floating fishing ground

Abstract

The present invention discloses a method for evaluating the aquaculture capacity of a submersible floating fish farm, comprising: Step S1, a step of calibrating the benchmark of the pixel points of cultured fish; Step S2, a step of measuring fish quantity data in layers; Step S3, a step of training a fish quantity estimation model; Step S4, a step of estimating the total fish quantity; The present invention can use an omnidirectional sonar to estimate the total fish quantity in the aquaculture net cage of the target submersible floating fish farm in a relatively simple and rapid data acquisition manner, so as to evaluate the aquaculture capacity of the aquaculture net cage of the target submersible floating fish farm, and has the advantage of accurate total fish quantity estimation result; Moreover, since the present invention only needs to use an omnidirectional sonar to obtain all the data required for the implementation method according to Step S2, it has the advantages of simplicity, convenience, high efficiency and speed, and can solve the problems of high requirements for data monitoring and acquisition of the aquaculture net cage of the target submersible floating fish farm, high price of monitoring equipment, complex operation and high cost in the prior art.

Description

Technical Field

[0001] The present invention relates to the estimation and measurement of fishery resources, and specifically to a method for evaluating the aquaculture capacity of a submersible and floating fishing ground. Background Art

[0002] At present, more than 10 countries in the world have developed deep-sea fishing ground aquaculture equipment, and more than a dozen deep-sea fishing ground equipment with excellent performance have been developed successively, which have been widely used in countries such as Norway, Finland, Sweden, the United States, Chile, Japan, Australia, Canada, the United Kingdom, and Denmark. Large-scale aquaculture fishing grounds have been established successively in Shandong, Guangdong, Fujian, Zhejiang and other places in China, and a series of aquaculture verifications have been carried out. The sea area with a water depth of more than 25m in China is vast, and the prospect of developing large-scale aquaculture fishing grounds is broad.

[0003] With the introduction of the national strategic document for actively developing the marine economy, the aquaculture industry faces the problem of moving from bays to the deep sea, and the requirements for the safety, automation and intelligent level of equipment have reached an unprecedented height. Especially in the South China Sea, there are a large number of floating HDPE cages distributed in this sea area. This type of cage has poor resistance to wind and waves, requires a large amount of manual labor, and has low production efficiency. The submersible and floating fishing ground large-scale deep-sea aquaculture equipment in the open sea of the South China Sea can withstand a 17-level typhoon, achieving the level of deep-sea aquaculture equipment to resist super typhoons, and can effectively solve the shortcoming of poor wind and wave resistance of traditional HDPE cage aquaculture. By carrying system equipment such as water quality monitoring and fish population monitoring, the aquaculture capacity of the submersible and floating fishing ground can be intelligently monitored, reducing manual operations, improving aquaculture efficiency, and realizing unmanned operation aquaculture.

[0004] It is difficult to exhaustively count the aquaculture capacity information data (such as quantity, distribution, etc.) of the submersible fishing ground, and it is not easy to form a specific, systematic and scientific aquaculture and management plan, which cannot provide sufficient technical support for the development of deep-sea aquaculture. With the rapid development of deep-sea aquaculture, the aquaculture water body and scale are constantly increasing, and the accurate evaluation of aquaculture capacity has become increasingly difficult. Traditional manual statistical methods often cannot meet the accuracy requirements. Secondly, the submersible and floating fishing ground has a large aquaculture water body and a large number of cultured fish, resulting in high manual statistical costs, and there may be underreporting or misreporting. In addition, factors such as weather and water temperature in the marine environment will also affect the evaluation of aquaculture capacity, further increasing the complexity of statistics. Therefore, accurately evaluating the aquaculture capacity of the submersible fishing ground is an important problem that needs to be solved urgently in the current deep-sea aquaculture industry. Summary of the Invention

[0005] The technical problem to be solved by the present invention is: to provide a method for evaluating the aquaculture capacity of a submersible and floating fishing ground.

[0006] To solve the above technical problems, the technical solutions adopted by the present invention are as follows:

[0007] A method for evaluating the aquaculture capacity of a submersible and floating fishing ground, characterized by comprising:

[0008] Step S1, the calibration step of the reference number of pixel points of cultured fish, which is:

[0009] Calibrate the target submersible and floating fishing ground aquaculture cage with an omnidirectional sonar to obtain the reference number of pixel points of a single fish, which represents the average number of pixel points of a fish in the sonar scan image scanned by the omnidirectional sonar.

[0010] Among them, the omnidirectional sonar can be a sonar that can achieve 360° omnidirectional scanning at the same time, or a sonar that realizes 360° omnidirectional scanning by driving the scanning beam to rotate in the horizontal plane. When the latter is used, since the scanning beam will form a spiral scanning path, the vertical movement speed of the omnidirectional sonar cannot be too fast, so as not to fail to comprehensively scan the overall space of the target submersible and floating fishing ground aquaculture cage.

[0011] Step S2, the step of stratified measurement of fish quantity data, including:

[0012] Step S2-1, Place the omnidirectional sonar at the center position of the target submersible and floating fishing ground aquaculture cage, and drive the omnidirectional sonar to move uniformly along the vertical direction from the bottom of the aquaculture cage of the target submersible and floating fishing ground to the water surface, or drive the omnidirectional sonar to move uniformly from the water surface to the bottom of the aquaculture cage of the target submersible and floating fishing ground to complete an overall scan.

[0013] Step S2-2, Play back the data output by the omnidirectional sonar for an overall scan through its supporting software to convert it into a single sonar scan video. And, intercept a video frame from the single sonar scan video every N seconds as a sonar scan image (see Figure 4 ), to obtain M / N sonar scan images arranged according to the interception time, which are sequentially recorded as the sonar scan images of the first layer to the M / N layer. Among them, M is the duration of the single sonar scan video, that is, the time to complete an overall scan, N is the time required for the omnidirectional sonar to complete a 360° scan in the horizontal direction, and M / N is an integer obtained by directly removing the decimal places and rounding.

[0014] Step S2-3, Perform binarization processing on the sonar scan image and remove noise points to identify the effective pixel points representing the cultured fish in the aquaculture cage of the target submersible and floating fishing ground in each layer of the sonar scan image. And, according to the reference number of pixel points of a single fish calibrated in step S1, calculate the number of cultured fish contained in the sonar scan images of each layer to the M / N layer, which is recorded as the stratified fish quantity sequence.

[0015] Step S3, the step of training the fish quantity estimation model, including:

[0016] Step S3-1: Execute step S2 multiple times to obtain multiple sets of training data. Each set of training data includes: a hierarchical fish quantity sequence and the corresponding actual fish quantity, where the actual fish quantity is the actual quantity of the cultured fish in the target submersible fish farm culture net cage when step S2 is executed.

[0017] Step S3-2: Use the hierarchical fish quantity sequence as the input and the actual fish quantity as the output, and train the AutoGluon model with the training data obtained in step S3-1 to obtain a fish quantity estimation model. Thus, the non-linear mapping problem between the hierarchical fish quantity sequence, that is, the quantity of cultured fish from the first layer to the M / N layer identified through step S2, and the actual quantity of the cultured fish in the target submersible fish farm culture net cage can be solved.

[0018] Step S4: Total fish quantity estimation step, including:

[0019] Step S4-1: At the target evaluation time, obtain the corresponding hierarchical fish quantity sequence according to step S2.

[0020] Step S4-2: Input the hierarchical fish quantity sequence obtained in step S4-1 into the fish quantity estimation model to output the total fish quantity estimation value of the target submersible fish farm culture net cage at the target evaluation time.

[0021] Therefore, through steps S1 to S4 of the present invention, the total fish quantity in the target submersible fish farm culture net cage can be estimated by using an omnidirectional sonar in a relatively simple and fast data acquisition manner, so as to evaluate the culture capacity of the target submersible fish farm culture net cage, and has the advantage of accurate total fish quantity estimation result.

[0022] Moreover, since the present invention only needs to use an omnidirectional sonar to obtain all the data required by the implementation method according to step S2, and the requirements for scanning the omnidirectional sonar are relatively low, as long as it moves vertically at a uniform speed to scan the entire target submersible fish farm culture net cage, it has the advantages of simplicity, convenience, high efficiency and speed, and can solve the problems of high requirements for data monitoring and acquisition of the target submersible fish farm culture net cage, high price of monitoring equipment, complex operation and high cost in the prior art.

[0023] Preferably, in step S1, the method for calibrating the reference number of pixels of a single fish is:

[0024] Step S1-1: Refer to Figure 2 , divide the target submersible fish farm culture net cage into multiple annular regions with equal radial widths centered on its central position.

[0025] Step S1-2: Place the omnidirectional sonar at the center of the target floating fish farm cage, and continuously track and scan the calibration fish in the target floating fish farm cage until the calibration fish swims through each of the said circular areas;

[0026] Step S1-3: From all the sonar scan images obtained by the omnidirectional sonar scanning in Step S1-2, extract the total number of pixels in each circular area for the calibration fish.

[0027] Among them, since the calibration fish will swim through the same circular area during multiple scans, the total number of pixels in the area is the total number of pixels scanned when the calibration fish swims through the same circular area each time.

[0028] Among them, the calibration fish can be one fish and its body size is the average of the cultured fish in the target floating fish farm cage, and thus the total number of pixels in the area can be directly scanned; or, the calibration fish can also be multiple fish, and the total number of pixels in the area is the total number of pixels scanned when each fish swims through the same circular area the same number of times ÷ the number of fish.

[0029] Step S1-4: Perform an averaging process on the total number of pixels in each circular area, that is: the total number of pixels in the area ÷ the number of times the calibration fish is scanned in this circular area, to calibrate the number of pixels for a single fish corresponding to each circular area; thus, by using the averaging process for each circular area, the problem that fish swimming in different postures in different environmental areas will present different numbers of pixels in the sonar scan image is solved.

[0030] In the said Step S2-3, the method for calculating the hierarchical fish quantity sequence is as follows:

[0031] Step S2-3-1: From the identified valid pixel points, count the number of valid pixel points in each circular area of the sonar scan image for each layer.

[0032] Step S2-3-2: For the sonar scan image of each layer: Calculate the number of cultured fish in each circular area through the number of valid pixel points in the same circular area and the number of pixels for a single fish benchmark, and record the total number of the number of cultured fish in all circular areas as the number of cultured fish contained in the sonar scan image of this layer, thereby obtaining the said hierarchical fish quantity sequence.

[0033] Thus, through steps S1-1 to S1-4, the present invention realizes the zoning of the annular area of the target submersible fish farm culture cage on the horizontal plane, calibrates the reference number of single fish pixel points corresponding to each annular area, and then through steps S2-3-1 and S2-3-2, calculates the number of cultured fish in each annular area one by one according to the annular area, so as to obtain the number of cultured fish included in the sonar scan image of each layer, that is, the stratified fish quantity sequence. Therefore, it is possible to reduce the influence of the mutual occlusion of the cultured fish in the target submersible fish farm culture cage on the scanning result of the omnidirectional sonar, and further improve the accuracy of the estimation of the total fish quantity in the target submersible fish farm culture cage by the present invention.

[0034] Preferably: in step S1-1, the radial width δd of each annular area is taken as the average fish length of the cultured fish in the target submersible fish farm culture cage.

[0035] Preferably: Refer to Figure 3 , the omnidirectional sonar is loaded on the unmanned aerial vehicle by a vertical lifting drive device, and the unmanned aerial vehicle suspends the omnidirectional sonar to the center position of the target submersible fish farm culture cage, and the vertical lifting drive device drives the omnidirectional sonar to move up and down in the vertical direction.

[0036] Preferably: in step S3-1, multiple executions of step S2 meet the following conditions:

[0037] Condition 1: When step S2 is executed each time, the actual number of fish in the target submersible fish farm culture cage is different;

[0038] Condition 2: The calibration fish used for calibrating the reference number of single fish pixel points in step S1 is of the same species and has a similar growth cycle as the cultured fish in the target submersible fish farm culture cage when steps S2 and S4 are executed.

[0039] Preferably: in step S3-2, the AutoGluon model is trained with the following settings:

[0040] The AutoGluon model uses two-layer stacking; the algorithm model for training is determined by AutoGluon, and all the algorithms used are trained with default hyperparameters, without setting the training time, and the trained model and results are automatically saved by the algorithm. The model will independently try all possible algorithm submodels for inverting the actual quantity from the quantity sequence, and continuously adjust the parameter scale and value of the submodel, and finally use the submodel with the highest score or the superposition of two submodels to perform the fish population quantity estimation task.

[0041] The AutoGluon model selects the best_quality gear for training to obtain a better model through a longer training time. The experimental data is divided into training data and test data, and the fish quantity table data of each water layer is converted into a csv format file. Actual quantity represents the actual number of fish, which is the target data to be inverted, and Depth represents the number of fish at the corresponding depth.

[0042] After the AutoGluon model reads the training data using TabularDataset(), the label is set to Actual quantity, and the num_gpus parameter is set to 1, which enables the use of the graphics card GPU to train the model, thereby accelerating the training speed. Then, TabularPredictor.fit() is used to start the training.

[0043] The AutoGluon model can train two types of models, the default gear and the best quality gear, and choose the better one for use.

[0044] Compared with the prior art, the present invention has the following beneficial effects:

[0045] First, through steps S1 to S4, the present invention can use the omnidirectional sonar 2 to estimate the total fish quantity in the target floating fish farm cage 1 in a relatively simple and fast data acquisition manner, so as to evaluate the breeding capacity of the target floating fish farm cage 1, and has the advantage of accurate total fish quantity estimation results.

[0046] Moreover, since the present invention only needs to use the omnidirectional sonar 2 to obtain all the data required by the implementation method according to step S2, the requirements for scanning the omnidirectional sonar 2 are relatively low. As long as it moves vertically at a uniform speed and scans the entire target floating fish farm cage 1, it has the advantages of simplicity, convenience, high efficiency and speed, and can solve the problems of high requirements for data monitoring and acquisition of the target floating fish farm cage 1, expensive monitoring equipment, complex operation and high cost in the prior art.

[0047] Second, through steps S1-1 to S1-4, the present invention realizes the zoning of the annular area of the target submersible fish farm cage 1 on the horizontal plane, calibrates the single fish pixel point number benchmark corresponding to each annular area, and then through steps S2-3-1 and S2-3-2, calculates the number of cultured fish in each annular area one by one according to the annular area, so as to obtain the number of cultured fish included in the sonar scan image of each layer, that is, the hierarchical fish number sequence. Thus, it can reduce the influence of the mutual occlusion of the cultured fish in the target submersible fish farm cage 1 on the scanning result of the omnidirectional sonar 2, and further improve the accuracy of the total fish quantity estimation of the present invention for the target submersible fish farm cage 1. BRIEF DESCRIPTION OF THE DRAWINGS

[0048] The present invention will be further described in detail below with reference to the drawings and specific embodiments:

[0049] Figure 1 is a flowchart of the present invention;

[0050] Figure 2 is a schematic diagram of dividing the annular area in step S1-1 of the present invention;

[0051] Figure 3 is a schematic diagram of the arrangement of the omnidirectional sonar in the target submersible fish farm cage of the present invention;

[0052] Figure 4 is a schematic diagram of the sonar scan image of the present invention. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0053] The present invention will be described in detail below with reference to the embodiments and their accompanying drawings to help those skilled in the art better understand the inventive concept of the present invention. However, the protection scope of the claims of the present invention is not limited to the following embodiments. For those skilled in the art, all other embodiments obtained without creative labor under the premise of not departing from the inventive concept of the present invention belong to the protection scope of the present invention.

[0054] As Figures 1 to 4 shown, the present invention discloses a method for evaluating the culture capacity of a submersible fish farm, including:

[0055] Step S1, the step of calibrating the pixel point number benchmark of cultured fish, is:

[0056] Use the omnidirectional sonar 2 to calibrate the target submersible fish farm cage 1 to obtain the single fish pixel point number benchmark, which represents the average pixel point number of a fish in the sonar scan image scanned by the omnidirectional sonar 2;

[0057] Among them, the omnidirectional sonar 2 can be a sonar that can simultaneously achieve 360° omnidirectional scanning, or a sonar that achieves 360° omnidirectional scanning by driving the scanning beam to rotate in the horizontal plane. When the latter is used, since the scanning beam will form a spiral scanning path, the vertical movement speed of the omnidirectional sonar cannot be too fast, so as not to fail to comprehensively scan the overall space of the target submerged floating fish farm cage 1.

[0058] In the step S1, the method for calibrating the reference number of pixel points of a single fish is as follows:

[0059] Step S1-1, referring to Figure 2 , divide the target submerged floating fish farm cage 1 into a plurality of annular regions with equal radial widths centered on its central position;

[0060] Step S1-2, place the omnidirectional sonar 2 at the central position of the target submerged floating fish farm cage 1, and continuously track and scan the calibration fish in the target submerged floating fish farm cage 1 until the calibration fish swims through each of the annular regions;

[0061] Step S1-3, extract the total number of pixel points of the calibration fish in each annular region from all the sonar scan images scanned by the omnidirectional sonar 2 in step S1-2;

[0062] Among them, since the calibration fish will swim through the same annular region during multiple scans, the total number of pixel points in the region is the total number of pixel points scanned when the calibration fish swims through the same annular region each time.

[0063] Among them, the calibration fish can be a single fish and its body size is the average of the cultured fish in the target submerged floating fish farm cage 1, so that the total number of pixel points in the region can be directly scanned; or, the calibration fish can also be multiple fish, and the total number of pixel points in the region is the total number of pixel points scanned when each fish swims through the same annular region the same number of times ÷ the number of fish.

[0064] Step S1-4, perform an averaging process on the total number of pixel points in each annular region, that is: the total number of pixel points in the region ÷ the number of times the calibration fish is scanned in this annular region, so as to calibrate the reference number of pixel points of a single fish corresponding to each annular region; thus, by using the averaging method for each annular region, the problem that fish swimming in different postures in different environmental regions will present different numbers of pixel points in the sonar scan image is solved.

[0065] Preferably: in the step S1-1, the radial width δd of each annular region is preferably set to the average fish length of the cultured fish in the target submerged floating fish farm cage 1.

[0066] Step S2, the fish quantity data stratified measurement step, includes:

[0067] Step S2-1: Place the omnidirectional sonar 2 at the center of the target submersible fish farm cage 1, and drive the omnidirectional sonar 2 to move uniformly along the vertical direction from the bottom surface 1-1 of the target submersible fish farm cage 1 to the water surface, or drive the omnidirectional sonar 2 to move uniformly from the water surface along the vertical direction to the bottom surface 1-1 of the target submersible fish farm cage 1 to complete an overall scan;

[0068] Step S2-2: Play back the data output by the omnidirectional sonar 2 during an overall scan through its supporting software to convert it into a single sonar scan video. And, intercept one video frame from the single sonar scan video every N seconds as a sonar scan image (see Figure 4 ), to obtain M / N sonar scan images arranged in the order of interception time, which are sequentially recorded as the sonar scan images of the first layer to the M / N layer. Among them, M is the duration of the single sonar scan video, that is, the time to complete an overall scan, N is the time required for the omnidirectional sonar 2 to complete a 360° scan in the horizontal direction, and M / N is an integer obtained by directly removing the decimal part and rounding;

[0069] Step S2-3: Perform binarization processing on the sonar scan images and remove noise points to identify the effective pixel points representing the cultured fish in the target submersible fish farm cage 1 in each layer of the sonar scan images. And, based on the single fish pixel point number benchmark calibrated in step S1, calculate the number of cultured fish contained in the sonar scan images from the first layer to the M / N layer, which is recorded as the hierarchical fish number sequence;

[0070] In the said step S2-3, the method for calculating the hierarchical fish number sequence is as follows:

[0071] Step S2-3-1: From the identified effective pixel points, count the number of effective pixel points in each annular region of each layer of the sonar scan images;

[0072] Step S2-3-2: For each layer of the sonar scan images: Calculate the number of cultured fish in each annular region through the number of effective pixel points in the same annular region and the single fish pixel point number benchmark, and record the total number of the number of cultured fish in all annular regions as the number of cultured fish contained in the sonar scan image of this layer, thereby obtaining the said hierarchical fish number sequence.

[0073] Thus, through steps S1-1 to S1-4, the present invention realizes the zoning of the annular area of the target submersible floating fish farm cage 1 on the horizontal plane, calibrates the reference number of single fish pixels corresponding to each annular area, and then through steps S2-3-1 and S2-3-2, calculates the number of cultured fish in each annular area one by one according to the annular area, so as to obtain the number of cultured fish included in the sonar scan image of each layer, that is, the hierarchical fish quantity sequence. Thus, it is possible to reduce the influence of the mutual occlusion of the cultured fish in the target submersible floating fish farm cage 1 on the scanning result of the omnidirectional sonar 2, and further improve the accuracy of the estimation of the total fish quantity in the target submersible floating fish farm cage 1 by the present invention.

[0074] Preferably: Refer to Figure 3 , the omnidirectional sonar 2 is loaded on the unmanned aerial vehicle 4 through the vertical lifting drive device 3, and the unmanned aerial vehicle 4 hoists the omnidirectional sonar 2 to the center position of the target submersible floating fish farm cage 1, and the vertical lifting drive device 3 drives the omnidirectional sonar 2 to move up and down in the vertical direction.

[0075] Step S3, fish quantity estimation model training step, includes:

[0076] Step S3-1, execute step S2 multiple times to obtain multiple groups of training data. Each group of training data includes: the hierarchical fish quantity sequence and the corresponding actual fish quantity, where the actual fish quantity is the actual quantity of the cultured fish in the target submersible floating fish farm cage 1 when step S2 is executed;

[0077] Preferably: In step S3-1, executing step S2 multiple times meets the following conditions:

[0078] Condition 1: When step S2 is executed each time, the actual fish quantity in the target submersible floating fish farm cage 1 is different;

[0079] Condition 2: The calibration fish used for calibrating the reference number of single fish pixels in step S1 has the same species and a similar growth cycle as the cultured fish in the target submersible floating fish farm cage 1 when steps S2 and S4 are executed.

[0080] Step S3-2, using the hierarchical fish quantity sequence as the input and the actual fish quantity as the output, train the AutoGluon model with the training data obtained in step S3-1 to obtain a fish quantity estimation model. Thus, it is possible to solve the non-linear mapping problem between the hierarchical fish quantity sequence, that is, the number of cultured fish from the first layer to the M / N layer recognized through step S2, and the actual fish quantity of the cultured fish in the target submersible floating fish farm cage 1;

[0081] Preferably, in the step S3-2, the AutoGluon model is trained with the following settings:

[0082] The AutoGluon model uses a two-layer stack; the algorithm model for training is determined by AutoGluon. All algorithms used are trained with default hyperparameters, without setting the training time. The trained model and results are automatically saved by the algorithm. The model will independently try all possible algorithm submodels that can invert the actual quantity from the quantity sequence, and continuously adjust the parameter scale and values of the submodels. Finally, the fish population quantity estimation task is performed using the submodel with the highest score or the superposition of two submodels.

[0083] The AutoGluon model selects the best_quality gear for training to obtain a better model through a longer training time. The experimental data is divided into training data and test data, and the fish quantity table data for each water layer is converted into a csv format file. Actual quantity represents the actual number of fish, which is the target data to be inverted, and Depth represents the number of fish at the corresponding depth.

[0084] After the AutoGluon model reads the training data using TabularDataset(), the label is set to Actual quantity, and the num_gpus parameter is set to 1, which enables the use of the graphics card GPU to train the model, thereby accelerating the training speed. Then, TabularPredictor.fit() is used to start the training;

[0085] The AutoGluon model can train two models, the default gear model and the best quality gear model, and choose the better one for use.

[0086] Step S4, the total fish quantity estimation step, includes:

[0087] Step S4-1, at the target evaluation time, obtain the corresponding stratified fish quantity sequence according to step S2;

[0088] Step S4-2, input the stratified fish quantity sequence obtained in step S4-1 into the fish quantity estimation model to output the total fish quantity estimation value of the target submersible floating fish farm cage 1 at the target evaluation time.

[0089] Thus, through steps S1 to S4, the present invention can use the omnidirectional sonar 2 to estimate the total fish quantity in the target submersible floating fish farm cage 1 in a relatively simple and fast data acquisition manner, so as to evaluate the aquaculture capacity of the target submersible floating fish farm cage 1, and has the advantage of accurate total fish quantity estimation results;

[0090] Moreover, since the present invention only needs to use the omnidirectional sonar 2 to obtain all the data required by the implementation method according to step S2, and the requirements for scanning the omnidirectional sonar 2 are relatively low. As long as it moves vertically at a uniform speed to scan the complete target submerged floating fish farm cage 1, it has the advantages of simplicity, convenience, high efficiency and speed, and can solve the problems of high requirements for data monitoring and collection of the target submerged floating fish farm cage 1 in the prior art, expensive monitoring equipment, complex operation and high cost.

[0091] The present invention is not limited to the above specific embodiments. According to the above content, based on the common general knowledge and customary means in the art, without departing from the above basic technical idea of the present invention, the present invention can also make various other forms of equivalent modifications, substitutions or changes, all of which fall within the protection scope of the present invention.

Claims

1. A method for evaluating the aquaculture capacity of a submersible and floating fishing ground, characterized in that, Including: Step S1, the calibration step of the pixel number benchmark of cultured fish, which is: Calibrate the target submerged floating fish farm culture net cage (1) with an omnidirectional sonar (2) to obtain the pixel number benchmark of a single fish, which represents the average pixel number of a fish in the sonar scan image scanned by the omnidirectional sonar (2); Step S2, the hierarchical measurement step of fish quantity data, including: Step S2-1, Place the omnidirectional sonar (2) at the center position of the target submerged floating fish farm culture net cage (1), and drive the omnidirectional sonar (2) to move uniformly from the bottom surface (1-1) of the net cage of the target submerged floating fish farm culture net cage (1) to the water surface along the vertical direction, or drive the omnidirectional sonar (2) to move uniformly from the water surface to the bottom surface (1-1) of the net cage of the target submerged floating fish farm culture net cage (1) along the vertical direction to complete an overall scan; Step S2-2, Play back the data output by the omnidirectional sonar (2) for an overall scan through its supporting software to convert it into a single sonar scan video, and, intercept a video frame from the single sonar scan video every N seconds as a sonar scan image to obtain M / N sonar scan images arranged according to the interception time, which are sequentially recorded as the sonar scan images of the first layer to the M / N layer, where M is the duration of the single sonar scan video, that is, the time to complete an overall scan, and N is the time required for the omnidirectional sonar (2) to complete a 360° scan in the horizontal direction; Step S2-3, Perform binarization processing on the sonar scan image and remove noise points to identify the effective pixel points representing the cultured fish in the sonar scan image of each layer, and, based on the pixel number benchmark of a single fish calibrated in step S1, calculate the number of cultured fish contained in the sonar scan images of each layer to the M / N layer, which is recorded as the hierarchical fish quantity sequence; Step S3, the training step of the fish quantity estimation model, including: Step S3-1, Execute step S2 multiple times to obtain multiple groups of training data, each group of training data includes: the hierarchical fish quantity sequence and the corresponding actual fish quantity, where the actual fish quantity is the actual quantity of the cultured fish in the target submerged floating fish farm culture net cage (1) when step S2 is executed; Step S3-2, Use the hierarchical fish quantity sequence as the input and the actual fish quantity as the output, and train the AutoGluon model with the training data obtained in step S3-1 to obtain the fish quantity estimation model; Step S4, the total fish quantity estimation step, including: Step S4-1, At the target evaluation time, obtain the corresponding hierarchical fish quantity sequence according to step S2; Step S4-2, Input the hierarchical fish quantity sequence obtained in step S4-1 into the fish quantity estimation model to output the total fish quantity estimation value of the target submerged floating fish farm culture net cage (1) at the target evaluation time.

2. The method for evaluating the aquaculture capacity of the submersible floating fishing ground according to claim 1, wherein: In the said step S1, the method for calibrating the pixel number benchmark of a single fish is: Step S1-1, Divide the target submerged floating fish farm culture net cage (1) into multiple annular regions with the same radial width centered on its center position; Step S1-2: Place the omnidirectional sonar (2) at the center of the target floating fish farm cage (1), and continuously track and scan the calibration fish in the target floating fish farm cage (1) until the calibration fish swim through each of the said circular areas; Step S1-3: Extract the total number of pixels in each circular area of the calibration fish from all the sonar scan images scanned by the omnidirectional sonar (2) in Step S1-2; Step S1-4: Perform an averaging process on the total number of pixels in each circular area to calibrate the pixel count benchmark for a single fish corresponding to each circular area; In the said Step S2-3, the method for calculating the stratified fish quantity sequence is as follows: Step S2-3-1: From the identified valid pixel points, count the number of valid pixel points in each circular area of the sonar scan image of each layer; Step S2-3-2: For the sonar scan image of each layer: Calculate the number of farmed fish in each circular area through the number of valid pixel points in the same circular area and the pixel count benchmark for a single fish, and record the total number of the number of farmed fish in all circular areas as the number of farmed fish contained in the sonar scan image of this layer, thereby obtaining the said stratified fish quantity sequence.

3. The method for evaluating the aquaculture capacity of the submersible floating fishery according to claim 2, wherein: In the said Step S1-1, the radial width δd of each circular area is taken as the average fish length of the farmed fish in the target floating fish farm cage (1).

4. The submersible and floating fishery farming capacity evaluation method according to any one of claims 1 to 3, characterized in that: The omnidirectional sonar (2) is loaded on the unmanned aerial vehicle (4) through a vertical lifting drive device (3). The unmanned aerial vehicle (4) hoists the omnidirectional sonar (2) to the center position of the target floating fish farm cage (1), and the vertical lifting drive device (3) drives the omnidirectional sonar (2) to move up and down in the vertical direction.

5. The submersible floating fish farm culture capacity evaluation method according to any one of claims 1 to 3, characterized in that: In the said Step S3-1, multiple executions of Step S2 meet the following conditions: Condition 1: When Step S2 is executed each time, the actual number of fish in the target floating fish farm cage (1) is different; Condition 2: The calibration fish used for calibrating the pixel count benchmark for a single fish in Step S1 is of the same species and has a similar growth cycle as the farmed fish in the target floating fish farm cage (1) when Steps S2 and S4 are executed.

6. The method for evaluating the aquaculture capacity of a submersible floating fishery according to any one of claims 1 to 3, characterized in that: In the said Step S3-2, the AutoGluon model is trained with the following settings: The AutoGluon model uses two-layer stacking; The AutoGluon model selects the best_quality gear for training; After the AutoGluon model reads the training data using TabularDataset(), set the label to Actual quantity, set the num_gpus parameter to 1, and then start training using TabularPredictor.fit().

Images