A method and system for monitoring fish swarm trajectory in marine ranches based on PIT

Abstract

The invention discloses a method and system for monitoring fish swarm trajectories in a marine ranch based on PIT. The method comprises: obtaining PIT tag information of each fish of a target fish swarm within a monitoring period based on a PIT reader; generating a three-dimensional fish swarm trajectory map of the target fish swarm within the monitoring period according to all PIT tag information of the target fish swarm within the monitoring period; for each fish, generating a motion sequence of each fish according to a corresponding three-dimensional motion trajectory, and inputting the motion sequence into a motion trajectory correction model so that the motion trajectory correction model outputs a corrected motion sequence; generating a first corrected three-dimensional fish swarm trajectory map of the target fish swarm within the monitoring period according to each corrected motion sequence, generating a second corrected three-dimensional fish swarm trajectory map according to the first corrected three-dimensional fish swarm trajectory map and the three-dimensional fish swarm trajectory map, generating motion information of the target fish swarm, and feeding back the second corrected three-dimensional fish swarm trajectory map and the motion information of the target fish swarm to monitoring personnel.

Description

Technical Field

[0001] The present invention relates to the technical field of fish track monitoring, and in particular to a method and system for monitoring fish tracks in a marine ranch based on PIT. Background Art

[0002] Fish school monitoring in marine ranches is an important means to ensure the sustainable use of marine fishery resources and improve aquaculture efficiency. Fish school monitoring in marine ranches mainly includes monitoring of fish number and density, fish behavior, fish growth status and fish trajectory. In terms of fish trajectory monitoring, the main method is to track the movement trajectory of fish schools, study their migration patterns and ecological habits, and provide guidance for the protection and utilization of fishery resources. The existing methods for monitoring fish trajectories in marine ranches mainly use satellite remote sensing technology, using satellite-borne remote sensors to conduct regular or continuous observations of marine ranches, and extract fish information through image processing and analysis technology. However, satellite remote sensing technology is affected by weather factors such as clouds and sea fog when taking images, resulting in shooting clarity problems. In addition, this technology is difficult to identify small fish schools or individuals, resulting in the omission of small fish school or individual data during monitoring, resulting in obvious lack of monitoring accuracy in the application of satellite remote sensing technology in fish trajectory monitoring. Summary of the invention

[0003] The embodiment of the present invention provides a method and system for monitoring the trajectory of fish schools in an ocean ranch based on PIT, which can solve the problem of low accuracy of group monitoring and inability to achieve individual monitoring of fish schools by analyzing the movement trajectory of fish schools through shooting fish movement images using satellite remote sensing technology, improve the accuracy of monitoring the movement trajectory of target fish schools in ocean ranches, and provide more accurate and rich data support for monitoring personnel's biological research on target fish schools.

[0004] An embodiment of the present invention provides a method for monitoring fish swarm trajectories in a marine ranch based on PIT, comprising:

[0005] Obtaining PIT tag information of each fish in the target fish school during the monitoring period based on the PIT reader; wherein each fish in the target fish school is implanted with a PIT before monitoring, and when the fish implanted with the PIT passes through the PIT reader, the PIT reader identifies the PIT and generates PIT tag information; the PIT tag information includes: a PIT tag, a timestamp, a first longitude coordinate, a first latitude coordinate, and a first vertical depth;

[0006] According to all PIT tag information of the target fish school in the monitoring period, a three-dimensional fish school trajectory map of the target fish school in the monitoring period is generated; wherein the three-dimensional fish school trajectory map is composed of the three-dimensional motion trajectories of all fish in the target fish school, and the three-dimensional motion trajectory of each fish includes a plurality of three-dimensional space points, and the three-dimensional coordinates of each three-dimensional space point include: a second longitude coordinate, a second latitude coordinate, and a second vertical depth;

[0007] For each fish, generating a motion sequence of each fish according to the corresponding three-dimensional motion trajectory, and inputting the motion sequence into the motion trajectory correction model so that the motion trajectory correction model outputs a corrected motion sequence;

[0008] Generate a first corrected three-dimensional fish school trajectory map of the target fish school within the monitoring period according to each corrected motion sequence, and generate a second corrected three-dimensional fish school trajectory map according to the first corrected three-dimensional fish school trajectory map and the three-dimensional fish school trajectory map;

[0009] The movement information of the target fish school is generated according to the second corrected three-dimensional fish school trajectory map, and the second corrected three-dimensional fish school trajectory map and the movement information of the target fish school are fed back to the monitoring personnel; wherein the movement information includes: the migration direction of the fish school and the movement speed of the fish school during the monitoring period.

[0010] Furthermore, the construction of the motion trajectory correction model includes:

[0011] Acquire a number of historical motion sequence samples of fish; wherein each historical motion sequence sample includes a number of historical three-dimensional coordinate samples with continuous time stamps;

[0012] Constructing an initial motion trajectory correction model; wherein the initial motion trajectory correction model includes a classification module and a trajectory correction module; the trajectory correction module includes a plurality of trajectory correction sub-models, each trajectory correction sub-model corresponding to a fish;

[0013] Training the classification module of the initial motion trajectory correction model with each historical motion sequence sample until the classification module reaches a first preset convergence condition, obtaining the network parameters of the classification module when the first preset convergence condition is reached, and freezing the network parameters of the classification module;

[0014] For each trajectory correction sub-model, training is performed using historical movement sequence samples of the fish corresponding to the trajectory correction sub-model until the current trajectory correction sub-model reaches a second preset convergence condition, obtaining network parameters of the current trajectory correction sub-model when the second preset convergence condition is reached, and freezing the network parameters of the current trajectory correction sub-model;

[0015] The network parameters of the initial motion trajectory correction model are adjusted by freezing the network parameters of the classification module and the network parameters of each trajectory correction sub-model when frozen, so as to generate a motion trajectory correction model.

[0016] Furthermore, for each trajectory correction sub-model, training is performed using the historical motion sequence samples of the fish corresponding to the trajectory correction sub-model, including:

[0017] According to the timestamp sequence of the historical running sequence samples, the historical motion sequence samples are divided into a plurality of motion sequence sample segments; wherein each motion sequence sample segment includes motion samples under three consecutive timestamps;

[0018] The trajectory correction sub-model is trained with each motion sequence sample segment; wherein, when each motion sequence sample segment is used to train the trajectory correction sub-model, the operation of selecting a motion sample at a timestamp as output and the motion samples at the remaining timestamps as input to train the trajectory correction sub-model is repeatedly performed until all motion samples at each timestamp in the current motion sequence sample segment are selected as output, and the next motion sequence sample segment is used to train the trajectory correction sub-model.

[0019] Furthermore, generating a three-dimensional fish school trajectory map of the target fish school within the monitoring period according to all PIT tag information of the target fish school within the monitoring period includes:

[0020] All PIT tag information is classified according to the PIT tag to obtain the PIT tag information grouping of each fish in the target fish school;

[0021] For each PIT marking information group, generating a three-dimensional motion trajectory corresponding to each PIT marking information group according to the timestamp, the first longitude coordinate, the first latitude coordinate and the first vertical depth;

[0022] The fish of the target fish school are divided according to the PIT markers, and several three-dimensional motion trajectories corresponding to each fish are obtained;

[0023] Fitting a number of three-dimensional motion trajectories of each fish to generate a fitted three-dimensional motion trajectory of each fish;

[0024] A three-dimensional fish school trajectory map of the target fish school during the monitoring period is generated based on the fitted three-dimensional motion trajectory of each fish in the target fish school and the three-dimensional motion trajectory corresponding to all PIT tag information groups.

[0025] Furthermore, for each fish, generating a motion sequence of each fish according to the corresponding three-dimensional motion trajectory includes:

[0026] For each fish in the target fish school, select a number of three-dimensional space points from the corresponding three-dimensional motion trajectory to generate a three-dimensional space point set, and record the timestamp of each three-dimensional space point; wherein the three-dimensional space point set at least includes the three-dimensional space point generated by the PIT tag information, the starting point of the current three-dimensional motion trajectory, and the end point of the current three-dimensional motion trajectory;

[0027] A motion sequence is generated according to the three-dimensional space point set and the time stamp of each three-dimensional space point in the three-dimensional space point set.

[0028] Further, generating a first corrected three-dimensional fish school trajectory map of the target fish school within the monitoring period according to each corrected motion sequence, and generating a second corrected three-dimensional fish school trajectory map according to the first corrected three-dimensional fish school trajectory map and the three-dimensional fish school trajectory map, including:

[0029] For each fish in the target fish school, a corresponding corrected three-dimensional motion trajectory is generated according to the corrected motion sequence;

[0030] Fitting the corrected three-dimensional motion trajectories belonging to the same fish according to the PIT markers to obtain the fitted corrected three-dimensional motion trajectory of each fish;

[0031] Generate a first corrected three-dimensional fish school trajectory map according to the fitted corrected three-dimensional motion trajectory of each fish in the target fish school and all corrected three-dimensional motion trajectories;

[0032] Constructing a geometric space body corresponding to the movement area of ​​the target fish school within the monitoring period according to the first corrected three-dimensional fish school trajectory map;

[0033] Taking the geometric center of the geometric space body as the coordinate origin, establishing a spatial rectangular coordinate system, converting each three-dimensional spatial point in the first corrected three-dimensional fish school trajectory map into a three-dimensional coordinate corresponding to the spatial rectangular coordinate system, and generating a third corrected three-dimensional fish school trajectory map;

[0034] Projecting the third corrected three-dimensional fish school trajectory image in a spatial rectangular coordinate system to obtain a first projection of the third corrected three-dimensional fish school trajectory image on a horizontal plane, a second projection on a vertical plane, and a third projection on a lateral plane in the spatial rectangular coordinate system;

[0035] Convert each three-dimensional space point in the three-dimensional fish school trajectory map into a three-dimensional coordinate corresponding to a spatial rectangular coordinate system to generate a three-dimensional fish school trajectory map to be corrected;

[0036] Projecting the three-dimensional fish school trajectory map to be corrected in a spatial rectangular coordinate system to obtain a fourth projection of the three-dimensional fish school trajectory map to be corrected in a horizontal plane, a fifth projection of the vertical plane, and a sixth projection of the lateral plane in the spatial rectangular coordinate system;

[0037] The fourth projection is corrected according to the first projection, the fifth projection is corrected according to the second projection, and the sixth projection is corrected according to the third projection to generate a second corrected three-dimensional fish school trajectory map.

[0038] Furthermore, after constructing a geometric space body corresponding to the movement area of ​​the target fish school within the monitoring period according to the first corrected three-dimensional fish school trajectory map, the method further includes:

[0039] Obtaining the coordinate information of the marine ecological red line, and determining a two-dimensional section model of the marine ecological red line according to the coordinate information of the marine ecological red line;

[0040] If the geometric space body corresponding to the movement area of ​​the target fish school during the monitoring period intersects with the two-dimensional section model of the marine ecological red line, determine the timestamp corresponding to the corrected three-dimensional fish school trajectory image segment within the marine ecological red line;

[0041] According to the timestamp corresponding to the corrected three-dimensional fish school trajectory image segment within the marine ecological red line, the corresponding three-dimensional fish school trajectory image segment is obtained, the three-dimensional coordinates of the three-dimensional space points in all the three-dimensional fish school trajectory image segments are obtained, and the three-dimensional coordinates of the three-dimensional space points in all the three-dimensional fish school trajectory image segments are converted into three-dimensional coordinates corresponding to the spatial rectangular coordinate system, to generate a fourth corrected three-dimensional fish school trajectory map;

[0042] Projecting the fourth corrected three-dimensional fish school trajectory image in a spatial rectangular coordinate system to obtain a seventh projection of the fourth corrected three-dimensional fish school trajectory image in a horizontal plane, an eighth projection of the vertical plane, and a ninth projection of the lateral plane in the spatial rectangular coordinate system;

[0043] The method of correcting the fourth projection according to the first projection, correcting the fifth projection according to the second projection, and correcting the sixth projection according to the third projection to generate a second corrected three-dimensional fish school trajectory map includes:

[0044] The fourth projection is corrected according to the first projection and the seventh projection, the fifth projection is corrected according to the second projection and the eighth projection, and the sixth projection is corrected according to the third projection and the ninth projection to generate a second corrected three-dimensional fish school trajectory map.

[0045] Based on the above method embodiment, the present invention provides a corresponding system embodiment;

[0046] An embodiment of the present invention provides a PIT-based marine ranch fish track monitoring system, including: a data acquisition module, a three-dimensional track map generation module, a track correction module and a monitoring information feedback module;

[0047] The data acquisition module is used to acquire the PIT tag information of each fish in the target fish school during the monitoring period based on the PIT reader; wherein each fish in the target fish school is implanted with a PIT before monitoring, and when the fish implanted with the PIT passes through the PIT reader, the PIT reader identifies the PIT and generates PIT tag information; the PIT tag information includes: a PIT tag, a timestamp, a first longitude coordinate, a first latitude coordinate, and a first vertical depth;

[0048] The three-dimensional trajectory map generating module is used to generate a three-dimensional fish school trajectory map of the target fish school within the monitoring period according to all PIT tag information of the target fish school within the monitoring period; wherein the three-dimensional fish school trajectory map is composed of the three-dimensional motion trajectories of all fish in the target fish school, and the three-dimensional motion trajectory of each fish includes a plurality of three-dimensional space points, and the three-dimensional coordinates of each three-dimensional space point include: a second longitude coordinate, a second latitude coordinate, and a second vertical depth;

[0049] The trajectory correction module is used to generate a motion sequence of each fish according to the corresponding three-dimensional motion trajectory for each fish, and input the motion sequence into the motion trajectory correction model so that the motion trajectory correction model outputs a corrected motion sequence; generate a first corrected three-dimensional fish school trajectory map of the target fish school within the monitoring period according to each corrected motion sequence, and generate a second corrected three-dimensional fish school trajectory map according to the first corrected three-dimensional fish school trajectory map and the three-dimensional fish school trajectory map;

[0050] The monitoring information feedback module is used to generate the movement information of the target fish school based on the second corrected three-dimensional fish school trajectory map, and feed back the second corrected three-dimensional fish school trajectory map and the movement information of the target fish school to the monitoring personnel; wherein the movement information includes: the migration direction of the fish school and the movement speed of the fish school during the monitoring period.

[0051] Another embodiment of the present invention provides a terminal device, including a processor, a memory, and a computer program stored in the memory and configured to be executed by the processor, wherein when the processor executes the computer program, it implements the PIT-based marine ranch fish trajectory monitoring method described in the above-mentioned embodiment of the invention.

[0052] Another embodiment of the present invention provides a storage medium, which includes a stored computer program, wherein when the computer program is running, the device where the storage medium is located is controlled to execute the PIT-based marine ranch fish trajectory monitoring method described in the above-mentioned invention embodiment.

[0053] The following beneficial effects are achieved by implementing the present invention:

[0054] The present invention provides a method and system for monitoring fish school trajectories in a marine ranch based on a PIT. The method implants a PIT in each fish of a target fish school, and then combines a PIT reader to obtain the PIT marking information of each fish of the target fish school during a monitoring period; wherein the PIT marking information includes: a PIT mark, a timestamp, a first longitude coordinate, a first latitude coordinate and a first vertical depth; and then generates a three-dimensional fish school trajectory diagram of the target fish school during the monitoring period according to all the PIT marking information of the target fish school during the monitoring period; and monitors each fish in the target fish school by implanting the PIT, that is, individual monitoring of the target fish school is achieved, so as to solve the problems of low group monitoring accuracy and inability to achieve individual monitoring in the fish school caused by analyzing the fish school motion trajectory by shooting fish school motion images through satellite remote sensing technology, thereby improving the accuracy of monitoring the motion trajectory of the target fish school in the marine ranch. Furthermore, after the three-dimensional motion trajectory generated by the PIT marking information is corrected by the trajectory correction model, the three-dimensional fish trajectory map generated by the PIT marking information is corrected, thereby further improving the accuracy of the movement information of the target fish during the monitoring period provided by the corrected three-dimensional fish trajectory map. When the corrected three-dimensional fish trajectory map and its corresponding better auxiliary motion information are provided to the monitoring personnel, it can provide the monitoring personnel with more accurate and rich data support for the biological research of the target fish. BRIEF DESCRIPTION OF THE DRAWINGS

[0055] Figure 1 It is a flow chart of a method for monitoring fish swarm trajectories in a marine ranch based on PIT provided in one embodiment of the present invention.

[0056] Figure 2 It is a structural schematic diagram of a PIT-based marine ranch fish trajectory monitoring system provided by an embodiment of the present invention. DETAILED DESCRIPTION

[0057] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0058] In order to better understand the technical solution of the present invention, it is necessary to explain in advance that PIT (Passive Integrated Transponder) is a miniature electronic tag developed based on the radio frequency identification system. It is small in size and can be implanted into the animal body by injection or other means. Each PIT tag has a built-in unique identification code, which gives each animal a unique identification number. It can achieve wireless and contactless identification.

[0059] like Figure 1 As shown, a method for monitoring fish swarm trajectories in a marine ranch based on PIT is provided in one embodiment of the present invention, comprising:

[0060] Step S1: obtaining the PIT tag information of each fish in the target fish school during the monitoring period based on the PIT reader; wherein each fish in the target fish school is implanted with a PIT before monitoring, and when the fish implanted with the PIT passes through the PIT reader, the PIT reader identifies the PIT and generates PIT tag information; the PIT tag information includes: a PIT tag, a timestamp, a first longitude coordinate, a first latitude coordinate, and a first vertical depth;

[0061] Step S2: generating a three-dimensional fish school trajectory map of the target fish school during the monitoring period according to all PIT tag information of the target fish school during the monitoring period; wherein the three-dimensional fish school trajectory map is composed of the three-dimensional motion trajectories of all fish in the target fish school, and the three-dimensional motion trajectory of each fish includes a plurality of three-dimensional space points, and the three-dimensional coordinates of each three-dimensional space point include: a second longitude coordinate, a second latitude coordinate, and a second vertical depth;

[0062] Step S3: for each fish, generating a motion sequence of each fish according to the corresponding three-dimensional motion trajectory, and inputting the motion sequence into the motion trajectory correction model, so that the motion trajectory correction model outputs a corrected motion sequence;

[0063] Step S4: generating a first corrected three-dimensional fish school trajectory map of the target fish school within the monitoring period according to each corrected motion sequence, and generating a second corrected three-dimensional fish school trajectory map according to the first corrected three-dimensional fish school trajectory map and the three-dimensional fish school trajectory map;

[0064] Step S5: Generate the movement information of the target fish school according to the second corrected three-dimensional fish school trajectory map, and feed back the second corrected three-dimensional fish school trajectory map and the movement information of the target fish school to the monitoring personnel; wherein the movement information includes: the migration direction of the fish school and the movement speed of the fish school during the monitoring period.

[0065] For step S1, a unique PIT is implanted into each fish in the target fish school to be monitored in the marine ranch before monitoring by injection, and the identity information of each fish, such as release time, birth information and fish information, is bound by the PIT tag in the PIT. The target fish school injected with the PIT is placed in the marine ranch, and the PIT tag information of the target fish school is obtained based on the PIT reader during the monitoring period. The PIT reader is usually arranged in a certain interval in the marine ranch, and the PIT carried by each fish school in the marine ranch is activated by transmitting a radio frequency signal. When the fish implanted with the PIT passes through the signal range of the PIT reader, the PIT reader identifies the PIT tag of the passing fish and generates the PIT tag information corresponding to the PIT tag, which includes the PIT tag, the timestamp corresponding to the recording time, the longitude coordinates of the PIT reader used for recording (i.e., the first longitude coordinates), the latitude coordinates of the PIT reader used for recording (i.e., the first latitude coordinates), and the vertical depth of the PIT reader used for recording from the sea level of the marine ranch (i.e., the first vertical depth). This method of injecting PIT into all individual fish in the target fish school and then obtaining the PIT tag information of each individual fish through a PIT reader can achieve individual monitoring of the target fish school, solving the problem of low accuracy of group monitoring and inability to achieve individual monitoring in the fish school by taking fish school movement images and analyzing fish school movement trajectories through satellite remote sensing technology.

[0066] However, since the PIT is implanted in the fish, it is not possible to directly obtain the specific location information of the fish implanted with the PIT in the marine ranch. The location information obtained is actually the location information of the PIT reader that the fish passes by. Therefore, in order to further improve the accuracy of individual location recognition, the present invention is implemented through the above steps S2-S4.

[0067] For step S2, a unit fish school trajectory map of the target fish school within the monitoring period is generated according to all PIT marking information of the target fish school within the monitoring period. The generation method can be achieved by inputting the PIT marking information into the three-dimensional trajectory map generation software.

[0068] In a preferred embodiment, the method of generating a three-dimensional fish school trajectory map of the target fish school during the monitoring period based on all PIT tag information of the target fish school during the monitoring period includes: classifying all PIT tag information according to the PIT tags to obtain PIT tag information groups of each fish in the target fish school; for each PIT tag information group, generating a three-dimensional motion trajectory corresponding to each PIT tag information group according to the timestamp, the first longitude coordinate, the first latitude coordinate and the first vertical depth; dividing the fish of the target fish school according to the PIT tags to obtain several three-dimensional motion trajectories corresponding to each fish; fitting several three-dimensional motion trajectories of each fish to generate a fitted three-dimensional motion trajectory of each fish; generating a three-dimensional fish school trajectory map of the target fish school during the monitoring period based on the fitted three-dimensional motion trajectory of each fish in the target fish school and the three-dimensional motion trajectories corresponding to all PIT tag information groups.

[0069] Specifically, since the PIT tag uniquely identifies the identity code of a fish corresponding to the PIT tag, all PIT tag information collected during the monitoring period can be classified through the PIT tag to obtain a PIT tag information group corresponding to each PIT tag; wherein each group contains the PIT tag information under all timestamps corresponding to the PIT tag. For each PIT tag information group, the first longitude coordinate, the first latitude coordinate and the first depth coordinate in the PIT tag information are used as the three-dimensional coordinates of the three-dimensional space point of the three-dimensional motion trajectory of the fish corresponding to the PIT tag, and several three-dimensional space points are generated in the software, and then the three-dimensional space points are connected according to the time sequence recorded by the timestamp, that is, the three-dimensional motion trajectory corresponding to the PIT tag is formed; wherein, after the three-dimensional motion trajectory is generated, the three-dimensional coordinates of the three-dimensional space point selected on the three-dimensional motion trajectory are represented by the second longitude coordinate, the second latitude coordinate and the second vertical depth. Furthermore, the fish of the target fish school are divided according to the PIT marks to obtain the number of fish contained in the target fish school and several three-dimensional motion trajectories corresponding to each fish; several three-dimensional motion trajectories belonging to the same fish are fitted to obtain the fitted three-dimensional motion trajectory of each fish; and then a three-dimensional fish school trajectory map of the target fish school during the monitoring period is generated according to the fitted three-dimensional motion trajectory of each fish in the target fish school and the three-dimensional motion trajectories corresponding to all PIT marks. Through the three-dimensional fish school trajectory map, the activity trajectory, activity range, movement speed of each fish in each time period and the activity trajectory, activity range and movement speed of each fish in each time period of the target fish during the monitoring time period can be preliminarily analyzed.

[0070] For step S3, for each fish, the three-dimensional motion trajectory generated by the grouping of its corresponding PIT tag information can be further refined to make it close to the actual motion trajectory of the fish. In the present invention, the three-dimensional motion trajectory generated by the software is first converted into an expression form of a motion sequence, and then the motion sequence is corrected by a pre-trained motion trajectory correction model to obtain a corrected motion sequence.

[0071] In a preferred embodiment, for each fish, a motion sequence of each fish is generated according to the corresponding three-dimensional motion trajectory, including: for each fish in the target fish school, a number of three-dimensional space points are selected from the corresponding three-dimensional motion trajectory to generate a three-dimensional space point set, and the timestamp of each three-dimensional space point is recorded; wherein the three-dimensional space point set includes at least the three-dimensional space point generated by the PIT marking information, the starting point of the current three-dimensional motion trajectory and the end point of the current three-dimensional motion trajectory; and a motion sequence is generated according to the three-dimensional space point set and the timestamp of each three-dimensional space point in the three-dimensional space point set.

[0072] For the generation of the motion sequence corresponding to each fish, it is necessary to select several three-dimensional space points from the corresponding three-dimensional motion trajectory to generate a three-dimensional space point set; when taking three-dimensional space points on the three-dimensional motion trajectory, it is necessary to select the reference point for generating the three-dimensional motion trajectory, that is, the three-dimensional space points whose second longitude coordinate, second latitude coordinate and second vertical depth in the current three-dimensional motion trajectory are the same as the first longitude coordinate, first latitude coordinate and first vertical depth in the PIT tag information at the same timestamp, as well as the starting point of the current three-dimensional motion trajectory and the end point of the current three-dimensional motion trajectory. In addition, in order to make the motion sequence data richer and the correction result better, in addition to the above-mentioned required three-dimensional space points, several three-dimensional space points are randomly selected from the three-dimensional motion trajectory and added to the three-dimensional space point set. The motion sequence is generated according to the timestamps corresponding to the three-dimensional space points contained in the final three-dimensional space point set, that is, the three-dimensional space point set; exemplarily, the motion sequence is expressed as [(starting point timestamp, three-dimensional coordinates of the starting point three-dimensional space point), (timestamp 1, three-dimensional coordinates of the three-dimensional space point of timestamp 1), (timestamp 2, three-dimensional coordinates of the three-dimensional space point of timestamp 2), ..., (end point timestamp, three-dimensional coordinates of the end point three-dimensional space point)].

[0073] In a preferred embodiment, the construction of the motion trajectory correction model includes: obtaining historical motion sequence samples of several fish; wherein each historical motion sequence sample includes several historical three-dimensional coordinate samples of continuous timestamps; constructing an initial motion trajectory correction model; wherein the initial motion trajectory correction model includes a classification module and a trajectory correction module; the trajectory correction module includes several trajectory correction sub-models, each trajectory correction sub-model corresponding to a fish; training the classification module of the initial motion trajectory correction model with each historical motion sequence sample until the classification module reaches a first preset convergence condition, obtaining the network parameters of the classification module when the first preset convergence condition is reached, and freezing the network parameters of the classification module; for each trajectory correction sub-model, training the historical motion sequence samples of the fish corresponding to the trajectory correction sub-model until the current trajectory correction sub-model reaches a second preset convergence condition, obtaining the network parameters of the current trajectory correction sub-model when the second preset convergence condition is reached, and freezing the network parameters of the current trajectory correction sub-model; adjusting the network parameters of the initial motion trajectory correction model with the frozen network parameters of the classification module and the network parameters of each trajectory correction sub-model when frozen, to generate a motion trajectory correction model.

[0074] Specifically, firstly, several historical motion sequence samples of fish are obtained; wherein each historical motion sequence sample includes several historical three-dimensional coordinate samples with continuous timestamps. It should be noted that in order to enable the trained motion trajectory correction model to realize the motion trajectory correction of all fish contained in the marine ranch, the historical motion sequence samples obtained here need to include the historical motion sequence samples of the existing fish in the marine ranch and the known fish to be released into the marine ranch.

[0075] An initial motion trajectory correction model is constructed, which includes a classification module and a trajectory correction module. The trajectory correction module includes several trajectory correction sub-models; each trajectory correction sub-model corresponds to a fish; for example, if the historical motion sequence samples of five types of fish are currently obtained, the trajectory correction module includes five trajectory correction sub-models, and the number of trajectory correction sub-models corresponds to the number of fish. The training of the initial motion trajectory correction model is divided into two parts, namely, the training of the splitting module and the training of the trajectory correction module. For the training of the classification module, each historical motion sequence sample is used as input and the corresponding fish is used as output, and the classification module is trained so that the classification module extracts and learns the fish feature information in each historical motion sequence sample, and outputs the prediction result according to the extracted fish feature information, compares the prediction result with the fish of the corresponding motion sequence sample, determines the error and calculates the loss function, and adjusts the network parameters of the classification module according to the error until the loss function of the classification module is minimized or the maximum number of iterations is reached (that is, when the above-mentioned first preset convergence condition is reached), obtains the network parameters of the classification module when the first preset convergence condition is reached, and freezes the network parameters.

[0076] For the training of each trajectory correction sub-model in the trajectory correction module, the historical motion sequence samples corresponding to each fish are respectively input into the trajectory correction sub-model of the corresponding fish for training. For example, the data of fish a is input into trajectory correction sub-model a, the data of fish b is input into trajectory correction sub-model b, and the data of fish n is input into trajectory correction sub-model n. By adopting this method that one fish corresponds to one trajectory correction sub-model, when introducing a new fish, it is only necessary to add a trajectory correction sub-model to the trajectory correction module, and retrain the trajectory correction sub-model and the classification module, thereby improving the flexibility of the motion trajectory correction model. When the loss function of any trajectory correction sub-model is the minimum or reaches the maximum number of iterations (i.e., when the second preset convergence condition is reached), the network parameters of the trajectory correction sub-model that reaches the second preset convergence condition are obtained and frozen; for the remaining trajectory correction sub-models that still need to be trained, the training is continued until convergence and then the network parameters are frozen. The network parameters of the initial motion trajectory correction model are adjusted by freezing the network parameters of the classification module and the network parameters of each trajectory correction sub-model when frozen to generate a motion trajectory correction model.

[0077] In a preferred embodiment, for each trajectory correction sub-model, the trajectory correction sub-model is trained with the historical motion sequence samples of the fish corresponding to the trajectory correction sub-model, including: dividing the historical motion sequence samples into a plurality of motion sequence sample segments according to the timestamp order of the historical motion sequence samples; wherein each motion sequence sample segment includes motion samples at three consecutive timestamps; training the trajectory correction sub-model with each motion sequence sample segment; wherein, when training the trajectory correction sub-model with each motion sequence sample segment, repeatedly executing the operation of selecting the motion sample at one timestamp as output and the motion samples at the remaining timestamps as input, and training the trajectory correction sub-model, until the motion samples at each timestamp in the current motion sequence sample segment are all selected as output, and executing the training of the trajectory correction sub-model with the next motion sequence sample segment.

[0078] Specifically, for the training of each trajectory correction sub-model, the historical motion sequence samples are divided into several continuous sequence sample segments according to the timestamp order of the historical motion sequence samples; wherein each motion sequence sample segment includes motion samples under three consecutive timestamps. Exemplarily, if the historical motion sequence samples corresponding to the current type A fish are: [(timestamp a, coordinate A), (timestamp b, coordinate B), (timestamp c, coordinate C), (timestamp d, coordinate D), ..., (timestamp n, coordinate N)], then the obtained motion sequence sample segments are exemplarily [(timestamp a, coordinate A), (timestamp b, coordinate B), (timestamp c, coordinate C)], [(timestamp b, coordinate B), (timestamp c, coordinate C), (timestamp d, coordinate D)].

[0079] The trajectory correction sub-model is trained with each motion sequence sample segment. When the trajectory correction sub-model is trained with each motion sequence sample segment, for example, when the trajectory correction sub-model is trained with [(timestamp a, coordinate A), (timestamp b, coordinate B), (timestamp c, coordinate C)], (timestamp b, coordinate B) is selected as the output, then (timestamp a, coordinate A) and (timestamp c, coordinate C) are used as input, and the training model outputs the predicted coordinate B corresponding to timestamp b according to (timestamp a, coordinate A) and (timestamp c, coordinate C). The predicted coordinate B is made close to the original coordinate B by adjusting the network parameters. After completing this round of training, timestamp b is marked as the output, and then (timestamp c, coordinate C) is selected as the input. Output, take (timestamp a, coordinate A) and (timestamp b, coordinate B) as input, train the model according to (timestamp a, coordinate A) and (timestamp b, coordinate B) to output the predicted coordinate C corresponding to timestamp c, adjust the network parameters to make the predicted coordinate C close to the original coordinate C, after completing this round of training, mark timestamp c as output, similarly, take (timestamp a, coordinate A) as output, (timestamp b, coordinate B) and (timestamp c, coordinate C) as input, train the model according to (timestamp b, coordinate B) and (timestamp c, coordinate C) to output the predicted coordinate A corresponding to timestamp a, adjust the network parameters to make the predicted coordinate A close to the original coordinate A, after completing this round of training, mark timestamp a as output. At this point, the training and learning of the correlation between the timestamps of the motion sequence sample fragments is completed, and the training and learning of the next motion sequence sample fragments are continued. The steps are the same as the above example and will not be repeated here. After training each trajectory correction sub-model in this way, the trajectory correction sub-model is able to predict the coordinate information corresponding to the missing timestamps based on adjacent or interval timestamps. Then, when the motion sequence lacking fine data is input into the trained trajectory correction sub-model, the motion sequence data can be finely filled, so that the obtained corrected motion sequence is smoother and has richer three-dimensional space point data.

[0080] For step S4, in a preferred embodiment, a first corrected three-dimensional fish school trajectory map of the target fish school within the monitoring period is generated according to each corrected motion sequence, and a second corrected three-dimensional fish school trajectory map is generated according to the first corrected three-dimensional fish school trajectory map and the three-dimensional fish school trajectory map, including: for each fish in the target fish school, a corresponding corrected three-dimensional motion trajectory is generated according to the corrected motion sequence; according to the PIT mark, each corrected three-dimensional motion trajectory belonging to the same fish is fitted to obtain a fitted corrected three-dimensional motion trajectory of each fish; according to the fitted corrected three-dimensional motion trajectory of each fish in the target fish school and all corrected three-dimensional motion trajectories, a first corrected three-dimensional fish school trajectory map is generated; according to the first corrected three-dimensional fish school trajectory map, a geometric space body corresponding to the movement area of ​​the target fish school within the monitoring period is constructed; with the geometric center of the geometric space body as the coordinate origin, a spatial rectangular coordinate system is established to convert the first corrected three-dimensional fish school trajectory map into a fixed three-dimensional fish school trajectory map; Each three-dimensional spatial point in the fish track map is converted into a three-dimensional coordinate corresponding to a spatial rectangular coordinate system to generate a third corrected three-dimensional fish track map; the third corrected three-dimensional fish track map is projected in the spatial rectangular coordinate system to obtain a first projection of the third corrected three-dimensional fish track map on the horizontal plane, a second projection on the vertical plane and a third projection on the lateral plane in the spatial rectangular coordinate system; each three-dimensional spatial point in the three-dimensional fish track map is converted into a three-dimensional coordinate corresponding to a spatial rectangular coordinate system to generate a three-dimensional fish track map to be corrected; the three-dimensional fish track map to be corrected is projected in the spatial rectangular coordinate system to obtain a fourth projection of the horizontal plane, a fifth projection of the vertical plane and a sixth projection of the lateral plane in the spatial rectangular coordinate system; the fourth projection is corrected according to the first projection, the fifth projection is corrected according to the second projection, and the sixth projection is corrected according to the third projection to generate a second corrected three-dimensional fish track map.

[0081] Specifically, after obtaining each corrected motion sequence, for each fish in the target fish school, a corresponding corrected three-dimensional motion trajectory is generated according to the corrected motion sequence, and the generation method can also be generated in software. According to the PIT mark, each corrected three-dimensional motion trajectory belonging to the same fish is fitted to obtain a fitted corrected three-dimensional motion trajectory of each fish; a first corrected three-dimensional fish school trajectory map is generated according to the fitted corrected three-dimensional motion trajectory of each fish in the target fish school and all corrected three-dimensional motion trajectories. After determining the first corrected three-dimensional fish school trajectory map, a geometric space body corresponding to the movement area of ​​the target fish school within the monitoring period is constructed based on the first corrected three-dimensional fish school trajectory map; wherein, the geometric space body needs to completely wrap all trajectory curves in the first corrected three-dimensional fish school trajectory map. Preferably, the geometric space body can be a cubic structure to facilitate obtaining the geometric center of the geometric space body. After determining the geometric center of the geometric space body, a spatial rectangular coordinate system is established with the geometric center as the coordinate origin, and the coordinates of each three-dimensional space point in the first corrected three-dimensional fish school trajectory map are converted from longitude, latitude and depth to the three-dimensional coordinate representation corresponding to the spatial rectangular coordinate system. The conversion method can be realized based on the existing longitude and latitude coordinates converted into two-dimensional plane coordinates. The deviation in the vertical direction is small and can be ignored in the conversion. The third corrected three-dimensional fish school trajectory map is generated according to the converted three-dimensional coordinates; the third corrected three-dimensional fish school trajectory map is projected in the spatial rectangular coordinate system to obtain the first projection of the third corrected three-dimensional fish school trajectory map on the horizontal plane, the second projection on the vertical plane and the third projection on the side plane in the spatial rectangular coordinate system. Similarly, the same conversion operation is performed on the three-dimensional fish school trajectory map to generate the three-dimensional fish school trajectory map to be corrected. The three-dimensional fish school trajectory map to be corrected is projected in a spatial rectangular coordinate system to obtain a fourth projection of the horizontal plane, a fifth projection of the vertical plane, and a sixth projection of the lateral plane of the three-dimensional fish school trajectory map to be corrected in the spatial rectangular coordinate system; the fourth projection is corrected according to the first projection, the fifth projection is corrected according to the second projection, and the sixth projection is corrected according to the third projection to generate a second corrected three-dimensional fish school trajectory map.

[0082] In a preferred embodiment, after constructing a geometric space body corresponding to the movement area of ​​the target fish school during the monitoring period according to the first corrected three-dimensional fish school trajectory map, it also includes: obtaining coordinate information of the marine ecological red line, and determining a two-dimensional section model of the marine ecological red line according to the coordinate information of the marine ecological red line; if the geometric space body corresponding to the movement area of ​​the target fish school during the monitoring period is interspersed with the two-dimensional section model of the marine ecological red line, determining the timestamp corresponding to the corrected three-dimensional fish school trajectory map segment within the marine ecological red line; obtaining the corresponding three-dimensional fish school trajectory map segment according to the timestamp corresponding to the corrected three-dimensional fish school trajectory map segment within the marine ecological red line, obtaining the three-dimensional coordinates of the three-dimensional spatial points in all the three-dimensional fish school trajectory map segments, and The three-dimensional coordinates of the spatial point are converted into the three-dimensional coordinates corresponding to the spatial rectangular coordinate system to generate a fourth corrected three-dimensional fish school trajectory map; the fourth corrected three-dimensional fish school trajectory map is projected in the spatial rectangular coordinate system to obtain the seventh projection of the horizontal plane, the eighth projection of the vertical plane and the ninth projection of the lateral plane of the fourth corrected three-dimensional fish school trajectory map in the spatial rectangular coordinate system; the fourth projection is corrected according to the first projection, the fifth projection is corrected according to the second projection, and the sixth projection is corrected according to the third projection to generate the second corrected three-dimensional fish school trajectory map, including: correcting the fourth projection according to the first projection and the seventh projection, correcting the fifth projection according to the second projection and the eighth projection, and correcting the sixth projection according to the third projection and the ninth projection to generate the second corrected three-dimensional fish school trajectory map.

[0083] Specifically, it should be noted that the marine ecological red line refers to the boundary line that demarcates important marine ecological functional areas, ecologically sensitive areas, and ecologically fragile areas as key control areas in order to maintain marine ecological health and ecological security. Therefore, when monitoring target fish schools, it is also necessary to consider the key monitoring of fish schools within the marine ecological red line. First, obtain the coordinate information of the marine ecological red line, and determine the two-dimensional section model of the marine ecological red line based on the coordinate information of the marine ecological red line. If the geometric space body corresponding to the movement area of ​​the target fish school during the monitoring period is interspersed with the two-dimensional section model of the marine ecological red line, that is, the movement area of ​​the target fish school during the monitoring period involves the marine ecological red line control area, it is necessary to determine the timestamp corresponding to the corrected three-dimensional fish school trajectory image segment within the marine ecological red line according to the marine ecological red line, obtain the corresponding three-dimensional fish school trajectory image segment according to the timestamp corresponding to the corrected three-dimensional fish school trajectory image segment within the marine ecological red line, obtain the three-dimensional coordinates of the three-dimensional space points in all three-dimensional fish school trajectory image segments, and convert the three-dimensional coordinates of the three-dimensional space points in all three-dimensional fish school trajectory image segments into three-dimensional coordinates corresponding to the spatial rectangular coordinate system to generate a fourth corrected three-dimensional fish school trajectory map, project the fourth corrected three-dimensional fish school trajectory map in the spatial rectangular coordinate system, and obtain the seventh projection of the horizontal plane, the eighth projection of the vertical plane, and the ninth projection of the lateral plane of the fourth corrected three-dimensional fish school trajectory map in the spatial rectangular coordinate system; then, the three-dimensional fish school trajectory map is corrected by combining the projection of the fourth corrected three-dimensional fish school trajectory map in the spatial rectangular coordinate system and the first corrected three-dimensional fish school trajectory map to generate a second corrected three-dimensional fish school trajectory map.

[0084] For step S5, the movement information of the target fish school is generated according to the second revised three-dimensional fish school trajectory map, and the movement information may include the migration direction of the fish school, that is, the movement direction of the target fish school is determined according to the latest timestamp of the second revised three-dimensional fish school trajectory map, and the movement speed of the fish school between each timestamp during the monitoring period. This information is fed back to the monitoring personnel, who can study the movement habits and movement areas of the target fish school during the monitoring period according to the second revised three-dimensional fish school trajectory map, and then infer the water resource content and other data corresponding to the movement area, so as to provide more accurate and rich data support for the biological research of the target fish school by the monitoring personnel.

[0085] Based on the above method embodiments, the present invention provides corresponding system embodiments.

[0086] like Figure 2 As shown, an embodiment of the present invention provides a PIT-based marine ranch fish track monitoring system, including: a data acquisition module, a three-dimensional track map generation module, a track correction module and a monitoring information feedback module;

[0087] The data acquisition module is used to acquire the PIT tag information of each fish in the target fish school during the monitoring period based on the PIT reader; wherein each fish in the target fish school is implanted with a PIT before monitoring, and when the fish implanted with the PIT passes through the PIT reader, the PIT reader identifies the PIT and generates PIT tag information; the PIT tag information includes: a PIT tag, a timestamp, a first longitude coordinate, a first latitude coordinate, and a first vertical depth;

[0088] The three-dimensional trajectory map generating module is used to generate a three-dimensional fish school trajectory map of the target fish school within the monitoring period according to all PIT tag information of the target fish school within the monitoring period; wherein the three-dimensional fish school trajectory map is composed of the three-dimensional motion trajectories of all fish in the target fish school, and the three-dimensional motion trajectory of each fish includes a plurality of three-dimensional space points, and the three-dimensional coordinates of each three-dimensional space point include: a second longitude coordinate, a second latitude coordinate, and a second vertical depth;

[0089] The trajectory correction module is used to generate a motion sequence of each fish according to the corresponding three-dimensional motion trajectory for each fish, and input the motion sequence into the motion trajectory correction model so that the motion trajectory correction model outputs a corrected motion sequence; generate a first corrected three-dimensional fish school trajectory map of the target fish school within the monitoring period according to each corrected motion sequence, and generate a second corrected three-dimensional fish school trajectory map according to the first corrected three-dimensional fish school trajectory map and the three-dimensional fish school trajectory map;

[0090] The monitoring information feedback module is used to generate the movement information of the target fish school based on the second corrected three-dimensional fish school trajectory map, and feed back the second corrected three-dimensional fish school trajectory map and the movement information of the target fish school to the monitoring personnel; wherein the movement information includes: the migration direction of the fish school and the movement speed of the fish school during the monitoring period.

[0091] It should be noted that the system embodiment described above is merely illustrative, wherein the units described as separate components may or may not be physically separated, and the components displayed as units may or may not be physical units, that is, they may be located in one place, or they may be distributed on multiple network units. Some or all of the modules may be selected according to actual needs to achieve the purpose of the scheme of this embodiment. In addition, in the drawings of the system embodiment provided by the present invention, the connection relationship between the modules indicates that there is a communication connection between them, which may be specifically implemented as one or more communication buses or signal lines. A person of ordinary skill in the art may understand and implement it without paying any creative effort.

[0092] Those skilled in the art can clearly understand that for the sake of convenience and brevity, the specific working process of the system described above can refer to the corresponding process in the aforementioned method embodiment, and will not be repeated here.

[0093] Based on the above method item embodiments, the present invention provides corresponding terminal device item embodiments.

[0094] An embodiment of the present invention provides a terminal device, including a processor, a memory, and a computer program stored in the memory and configured to be executed by the processor, wherein when the processor executes the computer program, a PIT-based marine ranch fish trajectory monitoring method as described in any one of the present invention is implemented.

[0095] The terminal device may be a computing device such as a desktop computer, a notebook, a PDA, a cloud server, etc. The terminal device may include, but is not limited to, a processor and a memory.

[0096] The processor may be a central processing unit (CPU), other general-purpose processors, digital signal processors (DSP), application-specific integrated circuits (ASIC), field-programmable gate arrays (FPGA) or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc. The general-purpose processor may be a microprocessor or any conventional processor, etc. The processor is the control center of the terminal device, and uses various interfaces and lines to connect various parts of the entire terminal device.

[0097] The memory can be used to store the computer program, and the processor realizes various functions of the terminal device by running or executing the computer program stored in the memory and calling the data stored in the memory. The memory can mainly include a program storage area and a data storage area, wherein the program storage area can store an operating system, at least one application required for a function, etc.; the data storage area can store data created according to the use of the mobile phone, etc. In addition, the memory can include a high-speed random access memory, and can also include a non-volatile memory, such as a hard disk, a memory, a plug-in hard disk, a smart memory card (Smart Media Card, SMC), a secure digital (Secure Digital, SD) card, a flash card (Flash Card), at least one disk storage device, a flash memory device, or other volatile solid-state storage devices.

[0098] Based on the above method item embodiments, the present invention provides a corresponding storage medium item embodiment.

[0099] An embodiment of the present invention provides a storage medium, which includes a stored computer program, wherein when the computer program is running, the device where the storage medium is located is controlled to execute a PIT-based marine ranch fish trajectory monitoring method as described in any one of the present inventions.

[0100] The storage medium is a computer-readable storage medium, and the computer program is stored in the computer-readable storage medium. When the computer program is executed by the processor, the steps of each of the above-mentioned method embodiments can be implemented. Among them, the computer program includes computer program code, and the computer program code can be in source code form, object code form, executable file or some intermediate form, etc. The computer-readable medium may include: any entity or device capable of carrying the computer program code, recording medium, USB flash drive, mobile hard disk, magnetic disk, optical disk, computer memory, read-only memory (ROM, Read-Only Memory), random access memory (RAM, Random Access Memory), electrical carrier signal, telecommunication signal and software distribution medium, etc.

[0101] The above is a preferred embodiment of the present invention. It should be pointed out that a person skilled in the art can make several improvements and modifications without departing from the principle of the present invention. These improvements and modifications are also considered to be within the scope of protection of the present invention.

Claims

1. A method for monitoring fish swarm trajectories in marine ranches based on PIT, characterized in that: include: Obtaining PIT tag information of each fish in the target fish school during the monitoring period based on the PIT reader; wherein each fish in the target fish school is implanted with a PIT before monitoring, and when the fish implanted with the PIT passes through the PIT reader, the PIT reader identifies the PIT and generates PIT tag information; the PIT tag information includes: a PIT tag, a timestamp, a first longitude coordinate, a first latitude coordinate, and a first vertical depth; According to all PIT tag information of the target fish school in the monitoring period, a three-dimensional fish school trajectory map of the target fish school in the monitoring period is generated; wherein the three-dimensional fish school trajectory map is composed of the three-dimensional motion trajectories of all fish in the target fish school, and the three-dimensional motion trajectory of each fish includes a plurality of three-dimensional space points, and the three-dimensional coordinates of each three-dimensional space point include: a second longitude coordinate, a second latitude coordinate, and a second vertical depth; For each fish, generating a motion sequence of each fish according to the corresponding three-dimensional motion trajectory, and inputting the motion sequence into the motion trajectory correction model so that the motion trajectory correction model outputs a corrected motion sequence; Generate a first corrected three-dimensional fish school trajectory map of the target fish school within the monitoring period according to each corrected motion sequence, and generate a second corrected three-dimensional fish school trajectory map according to the first corrected three-dimensional fish school trajectory map and the three-dimensional fish school trajectory map; Generate the movement information of the target fish school according to the second corrected three-dimensional fish school trajectory map, and feed back the second corrected three-dimensional fish school trajectory map and the movement information of the target fish school to the monitoring personnel; wherein the movement information includes: the migration direction of the fish school and the movement speed of the fish school during the monitoring period; The construction of the motion trajectory correction model includes: Acquire a number of historical motion sequence samples of fish; wherein each historical motion sequence sample includes a number of historical three-dimensional coordinate samples with continuous time stamps; Constructing an initial motion trajectory correction model; wherein the initial motion trajectory correction model includes a classification module and a trajectory correction module; the trajectory correction module includes a plurality of trajectory correction sub-models, each trajectory correction sub-model corresponding to a fish; Training the classification module of the initial motion trajectory correction model with each historical motion sequence sample until the classification module reaches a first preset convergence condition, obtaining the network parameters of the classification module when the first preset convergence condition is reached, and freezing the network parameters of the classification module; For each trajectory correction sub-model, training is performed using historical movement sequence samples of the fish corresponding to the trajectory correction sub-model until the current trajectory correction sub-model reaches a second preset convergence condition, obtaining network parameters of the current trajectory correction sub-model when the second preset convergence condition is reached, and freezing the network parameters of the current trajectory correction sub-model; The network parameters of the initial motion trajectory correction model are adjusted by freezing the network parameters of the classification module and the network parameters of each trajectory correction sub-model when frozen, so as to generate a motion trajectory correction model; The method of generating a three-dimensional fish school trajectory map of the target fish school within the monitoring period according to all PIT tag information of the target fish school within the monitoring period includes: All PIT tag information is classified according to the PIT tag to obtain the PIT tag information grouping of each fish in the target fish school; For each PIT marking information group, generating a three-dimensional motion trajectory corresponding to each PIT marking information group according to the timestamp, the first longitude coordinate, the first latitude coordinate and the first vertical depth; The fish of the target fish school are divided according to the PIT markers, and several three-dimensional motion trajectories corresponding to each fish are obtained; Fitting a number of three-dimensional motion trajectories of each fish to generate a fitted three-dimensional motion trajectory of each fish; A three-dimensional fish school trajectory map of the target fish school during the monitoring period is generated based on the fitted three-dimensional motion trajectory of each fish in the target fish school and the three-dimensional motion trajectory corresponding to all PIT tag information groups.

2. A method for monitoring fish swarm trajectories in marine ranches based on PIT as claimed in claim 1, characterized in that: For each trajectory correction sub-model, training is performed using the trajectory correction sub-model corresponding to the historical movement sequence samples of the fish, including: According to the timestamp sequence of the historical running sequence samples, the historical motion sequence samples are divided into a plurality of motion sequence sample segments; wherein each motion sequence sample segment includes motion samples under three consecutive timestamps; The trajectory correction sub-model is trained with each motion sequence sample segment; wherein, when each motion sequence sample segment is used to train the trajectory correction sub-model, the operation of selecting a motion sample at a timestamp as output and the motion samples at the remaining timestamps as input to train the trajectory correction sub-model is repeatedly performed until all motion samples at each timestamp in the current motion sequence sample segment are selected as output, and the next motion sequence sample segment is used to train the trajectory correction sub-model.

3. A method for monitoring fish swarm trajectories in marine ranches based on PIT as claimed in claim 1, characterized in that: For each fish, generating a motion sequence of each fish according to the corresponding three-dimensional motion trajectory includes: For each fish in the target fish school, select a number of three-dimensional space points from the corresponding three-dimensional motion trajectory to generate a three-dimensional space point set, and record the timestamp of each three-dimensional space point; wherein the three-dimensional space point set at least includes the three-dimensional space point generated by the PIT tag information, the starting point of the current three-dimensional motion trajectory, and the end point of the current three-dimensional motion trajectory; A motion sequence is generated according to the three-dimensional space point set and the time stamp of each three-dimensional space point in the three-dimensional space point set.

4. A method for monitoring fish swarm trajectories in marine ranches based on PIT as claimed in claim 3, characterized in that: Generating a first corrected three-dimensional fish school trajectory map of the target fish school within the monitoring period according to each corrected motion sequence, and generating a second corrected three-dimensional fish school trajectory map according to the first corrected three-dimensional fish school trajectory map and the three-dimensional fish school trajectory map, including: For each fish in the target fish school, a corresponding corrected three-dimensional motion trajectory is generated according to the corrected motion sequence; Fitting the corrected three-dimensional motion trajectories belonging to the same fish according to the PIT markers to obtain the fitted corrected three-dimensional motion trajectory of each fish; Generate a first corrected three-dimensional fish school trajectory map according to the fitted corrected three-dimensional motion trajectory of each fish in the target fish school and all corrected three-dimensional motion trajectories; Constructing a geometric space body corresponding to the movement area of ​​the target fish school within the monitoring period according to the first corrected three-dimensional fish school trajectory map; Taking the geometric center of the geometric space body as the coordinate origin, establishing a spatial rectangular coordinate system, converting each three-dimensional spatial point in the first corrected three-dimensional fish school trajectory map into a three-dimensional coordinate corresponding to the spatial rectangular coordinate system, and generating a third corrected three-dimensional fish school trajectory map; Projecting the third corrected three-dimensional fish school trajectory image in a spatial rectangular coordinate system to obtain a first projection of the third corrected three-dimensional fish school trajectory image on a horizontal plane, a second projection on a vertical plane, and a third projection on a lateral plane in the spatial rectangular coordinate system; Convert each three-dimensional space point in the three-dimensional fish school trajectory map into a three-dimensional coordinate corresponding to a spatial rectangular coordinate system to generate a three-dimensional fish school trajectory map to be corrected; Projecting the three-dimensional fish school trajectory map to be corrected in a spatial rectangular coordinate system to obtain a fourth projection of the three-dimensional fish school trajectory map to be corrected in a horizontal plane, a fifth projection of the vertical plane, and a sixth projection of the lateral plane in the spatial rectangular coordinate system; The fourth projection is corrected according to the first projection, the fifth projection is corrected according to the second projection, and the sixth projection is corrected according to the third projection to generate a second corrected three-dimensional fish school trajectory map.

5. A method for monitoring fish swarm trajectories in marine ranches based on PIT as claimed in claim 4, characterized in that: After constructing a geometric space body corresponding to the movement area of ​​the target fish school within the monitoring period according to the first corrected three-dimensional fish school trajectory map, the method further includes: Obtaining the coordinate information of the marine ecological red line, and determining a two-dimensional section model of the marine ecological red line according to the coordinate information of the marine ecological red line; If the geometric space body corresponding to the movement area of ​​the target fish school during the monitoring period intersects with the two-dimensional section model of the marine ecological red line, determine the timestamp corresponding to the corrected three-dimensional fish school trajectory image segment within the marine ecological red line; According to the timestamp corresponding to the corrected three-dimensional fish school trajectory image segment within the marine ecological red line, the corresponding three-dimensional fish school trajectory image segment is obtained, the three-dimensional coordinates of the three-dimensional space points in all the three-dimensional fish school trajectory image segments are obtained, and the three-dimensional coordinates of the three-dimensional space points in all the three-dimensional fish school trajectory image segments are converted into three-dimensional coordinates corresponding to the spatial rectangular coordinate system, to generate a fourth corrected three-dimensional fish school trajectory map; Projecting the fourth corrected three-dimensional fish school trajectory image in a spatial rectangular coordinate system to obtain a seventh projection of the fourth corrected three-dimensional fish school trajectory image in a horizontal plane, an eighth projection of the vertical plane, and a ninth projection of the lateral plane in the spatial rectangular coordinate system; The method of correcting the fourth projection according to the first projection, correcting the fifth projection according to the second projection, and correcting the sixth projection according to the third projection to generate a second corrected three-dimensional fish school trajectory map includes: The fourth projection is corrected according to the first projection and the seventh projection, the fifth projection is corrected according to the second projection and the eighth projection, and the sixth projection is corrected according to the third projection and the ninth projection to generate a second corrected three-dimensional fish school trajectory map.

6. A PIT-based marine ranch fish track monitoring system, the monitoring system implements the PIT-based marine ranch fish track monitoring method as claimed in claim 1, characterized in that: include: Data acquisition module, three-dimensional trajectory map generation module, trajectory correction module and monitoring information feedback module; The data acquisition module is used to acquire the PIT tag information of each fish in the target fish school during the monitoring period based on the PIT reader; wherein each fish in the target fish school is implanted with a PIT before monitoring, and when the fish implanted with the PIT passes through the PIT reader, the PIT reader identifies the PIT and generates PIT tag information; the PIT tag information includes: a PIT tag, a timestamp, a first longitude coordinate, a first latitude coordinate, and a first vertical depth; The three-dimensional trajectory map generating module is used to generate a three-dimensional fish school trajectory map of the target fish school within the monitoring period according to all PIT tag information of the target fish school within the monitoring period; wherein the three-dimensional fish school trajectory map is composed of the three-dimensional motion trajectories of all fish in the target fish school, and the three-dimensional motion trajectory of each fish includes a plurality of three-dimensional space points, and the three-dimensional coordinates of each three-dimensional space point include: a second longitude coordinate, a second latitude coordinate, and a second vertical depth; The trajectory correction module is used to generate a motion sequence of each fish according to the corresponding three-dimensional motion trajectory for each fish, and input the motion sequence into the motion trajectory correction model so that the motion trajectory correction model outputs a corrected motion sequence; generate a first corrected three-dimensional fish school trajectory map of the target fish school within the monitoring period according to each corrected motion sequence, and generate a second corrected three-dimensional fish school trajectory map according to the first corrected three-dimensional fish school trajectory map and the three-dimensional fish school trajectory map; The monitoring information feedback module is used to generate the movement information of the target fish school based on the second corrected three-dimensional fish school trajectory map, and feed back the second corrected three-dimensional fish school trajectory map and the movement information of the target fish school to the monitoring personnel; wherein the movement information includes: the migration direction of the fish school and the movement speed of the fish school during the monitoring period.

7. A terminal device, characterized in that: It includes a processor, a memory, and a computer program stored in the memory and configured to be executed by the processor, and when the processor executes the computer program, it implements a PIT-based marine ranch fish trajectory monitoring method as described in any one of claims 1 to 5.

8. A storage medium, characterized in that: The storage medium includes a stored computer program, wherein when the computer program is running, the device where the storage medium is located is controlled to execute the PIT-based marine ranch fish trajectory monitoring method as described in any one of claims 1 to 5.

Images