Large-model-based intelligent identification shrimp group living body health monitoring method

Abstract

The invention relates to the field of living body health monitoring, in particular to a method for intelligently recognizing the living body health of a shrimp group based on a large model, and the method comprises the steps: continuously obtaining the images of the shrimp group in a culture pond, and extracting the active distribution characteristics of the shrimp group; analyzing an active offset characterization value of the prawn group by combining the swimming speed of the prawn cluster, and marking the prawn group; in response to a marking result, analyzing the prawn group, including identifying a trajectory swimming discrete stage of the prawn group, and analyzing a plurality of prawn clusters in the trajectory swimming discrete stage; according to the method and the system, on the basis of real-time and continuous monitoring of the prawn clusters, the accuracy of health monitoring of the prawn clusters is improved, the prawn clusters are managed in a grading manner, and the prawn clusters are effectively protected. The overall influence of the prawn clusters and prawn groups possibly having health problems is reduced, and the overall health level of the prawn groups is guaranteed.

Description

TECHNICAL FIELD

[0001] The present application relates to the field of living body health monitoring, and in particular to a method for intelligent identification of shrimp group living body health monitoring based on a large model. BACKGROUND

[0002] With the continuous expansion of shrimp farming scale and the improvement of the degree of intensification, the problems such as frequent disease of shrimp group and difficult real-time monitoring of growth state are increasingly prominent, which seriously restricts the sustainable development of the industry. At the same time, in view of the characteristics of high-density intensive shrimp culture, such as high density, closed environment, high risk of disease, etc., a real-time, accurate and automated health monitoring system is needed to ensure the health of shrimp group and improve the efficiency of aquaculture.

[0003] Chinese patent application publication No. CN119646725A discloses a fish health condition monitoring device based on fish behavior analysis, relating to aquaculture technology, comprising a water surface unit, a water unit, a water bottom unit and a data processing center. The water surface unit, the water unit and the water bottom unit are respectively installed on the water surface, in the water and at the water bottom of the culture pond. The water surface unit, the water unit and the water bottom unit respectively collect real-time data of the water surface, in the water and at the water bottom of the culture pond. The data processing center receives the real-time data and processes the real-time data through a Stacking stacked learning model to judge the health condition of the fish group in the culture pond and the position where the abnormality occurs. When the fish group in a certain position in the judged area behaves abnormally, an abnormal alarm signal is generated. The application also discloses a fish health condition monitoring method based on fish behavior analysis. The application can realize real-time detection and early warning of the health condition of the fish group, effectively improving the risk resistance and economic benefits of the culture process.

[0004] However, the prior art still has the following problems, The traditional method for detecting the health of shrimp group mainly relies on manual inspection and sampling detection. The former relies on the relevant experience of the breeding personnel, and the latter has the problems of insufficient sample representativeness and long detection period, which cannot realize real-time and continuous monitoring of the shrimp group. At the same time, the shrimp bodies with health problems cannot be effectively treated in time. SUMMARY

[0005] Therefore, the present application provides a method for intelligent identification of shrimp group living body health monitoring based on a large model to overcome the problems in the prior art that the shrimp group cannot be monitored in real time and continuously, and the shrimp bodies with health problems cannot be effectively treated in time.

[0006] To achieve the above purpose, the present application provides a method for intelligent identification of shrimp group living body health monitoring based on a large model, which comprises: Continuously acquire images of the shrimp group in the culture pond to extract active distribution characteristics of the shrimp group, the active distribution characteristics including a swimming offset shrimp cluster density and a swimming depth variation amount of a corresponding shrimp cluster; According to the active distribution characteristics and the swimming speed of the shrimp cluster, an active offset characteristic value of the shrimp group is analyzed to label the shrimp group. In response to the labeling result, the shrimp group is analyzed, including, A trajectory swimming discrete stage of the shrimp group is identified, and a plurality of shrimp clusters in the trajectory swimming discrete stage are analyzed, including, acquiring respective branch trajectories corresponding to the trajectory swimming discrete stage, extracting swimming data of a corresponding minimum shrimp cluster in each branch trajectory, calling a horizontal shortest distance of the corresponding minimum shrimp cluster from an obstacle in the culture pond and a stacking amount of the minimum shrimp cluster to evaluate a health abnormality characteristic coefficient of the minimum shrimp cluster to determine whether to clean the minimum shrimp cluster. Swimming trajectories of a plurality of remaining shrimp clusters are acquired, trajectory tendency values of the respective remaining shrimp clusters are determined, and the trajectory tendency values are matched with a trajectory tendency reference range to determine whether to isolate the remaining shrimp clusters.

[0007] Further, the process of analyzing the active offset characteristic value of the shrimp group according to the active distribution characteristics and the swimming speed of the shrimp cluster includes, a ratio of the swimming offset shrimp cluster density to a shrimp cluster density threshold value and a ratio of the swimming depth variation amount of the corresponding shrimp cluster to a swimming depth variation threshold value are summed as a first active offset feature; a ratio of the swimming speed threshold value to the swimming speed of the shrimp cluster is taken as a second active offset feature; the first active offset feature and the second active offset feature are weighted and summed to determine the active offset characteristic value.

[0008] Further, the labeling of the shrimp group includes, If the active offset characteristic value of the shrimp group is greater than or equal to an active offset characteristic threshold value, the shrimp group is labeled with a high offset degree label.

[0009] Further, in response to the labeling result, the shrimp group is analyzed, including, If any shrimp group is labeled with a high offset degree label, the shrimp group is analyzed.

[0010] Further, the process of identifying the trajectory swimming discrete stage of the shrimp group includes, shrimp group images are called to identify a swimming trajectory of the shrimp group; If the swimming trajectory meets trajectory discrete conditions, it is determined that the shrimp group enters a trajectory swimming discrete stage. The trajectory meets a trajectory forming condition, which includes two or more swimming trajectories appearing and being maintained for a predetermined time length.

[0011] Further, the process of evaluating the health abnormality representation coefficient of the minimum shrimp cluster includes, a ratio of the horizontal minimum distance threshold value to the horizontal minimum distance of the minimum shrimp cluster from the obstacle in the aquaculture pond as a first health abnormality feature; a ratio of the stacking amount of the minimum shrimp cluster to the stacking amount threshold value as a second health abnormality feature; a sum of the first health abnormality feature and the second health abnormality feature as the health abnormality representation coefficient.

[0012] Further, the process of determining whether to clean up the minimum shrimp cluster includes, if the health abnormality representation coefficient of the minimum shrimp cluster is greater than or equal to a health abnormality representation coefficient threshold value, determining to clean up the minimum shrimp cluster.

[0013] Further, the process of determining the trajectory trend value of each of the remaining shrimp clusters includes, generating a plurality of vectors corresponding to the swimming trajectories of a plurality of remaining shrimp clusters, respectively; selecting a plurality of remaining target shrimp clusters for which the vector directions of the plurality of vectors remain unchanged within a predetermined time; determining a distance shortening amount of each of the remaining target shrimp clusters from the minimum shrimp cluster within the predetermined time; the distance shortening amount as the trajectory trend value.

[0014] Further, the process of determining whether to isolate the remaining shrimp cluster includes, if the trajectory trend value of any of the remaining target shrimp clusters is not within a trajectory trend reference range, determining to isolate the remaining target shrimp cluster; wherein the trajectory trend reference range is pre-set.

[0015] Further, it further includes setting a response prompt signal, in response to the minimum shrimp cluster being cleaned up or the remaining target shrimp cluster being isolated, the prompt signal is issued.

[0016] Compared with the prior art, the present application can improve the accuracy of health monitoring of the shrimp group on the basis of real-time and continuous monitoring of the shrimp group, and can reduce the influence of the shrimp cluster that may have health problems on the whole shrimp group through hierarchical management of the shrimp cluster, thereby ensuring the health level of the whole shrimp group.

[0017] In particular, based on the factors such as the decline of physiological function and the decrease of environmental adaptability of the sick shrimp that has health problems, the sick shrimp separates from the shrimp group and forms a separate swimming track, the swimming active distribution of the shrimp group in the culture pond is considered, the density of the shrimp cluster that deviates from the swimming of the shrimp group reflects the damage degree of the sick shrimp to the whole shrimp group, and compared with the normal shrimp group that is uniformly distributed in the upper layer or the flow area of the water body, the sick shrimp usually gathers in the deep places such as the bottom and the corner of the culture pond, so the change amount of the swimming depth of the shrimp cluster can reflect the degree of physical exhaustion and the degree of decline of swimming ability of the sick shrimp, and the swimming speed can directly reflect the vitality level of the shrimp cluster, therefore, the present application can comprehensively capture the activity and health status of the whole shrimp group by combining the above features, the active deviation characteristic value is used to represent the severity of the dispersion of the swimming track of the shrimp group and the deviation degree of the swimming track of the dispersed shrimp cluster, data support is provided for subsequent marking of the shrimp group, and the shrimp group is adaptively analyzed, the present application can improve the accuracy of health monitoring of the shrimp group on the basis of real-time and continuous monitoring of the shrimp group, and can reduce the influence of the shrimp cluster that may have health problems on the whole shrimp group through hierarchical management of the shrimp cluster, thereby ensuring the health level of the whole shrimp group.

[0018] Especially, the application carries out differential treatment monitoring for shrimp groups provided with high offset degree labels, focuses on analyzing the discrete swimming trajectories of the whole shrimp group, locks high-risk weak areas, realizes dynamic positioning of weak shrimp clusters and preferentially carries out waste cleaning and other treatments. Weak shrimps are unable to resist water flow impact due to their physical exhaustion or nerve damage, and are easily pushed to the vicinity of obstacles, so that the horizontal distance between them and the obstacles in the culture pond is shortened. At the same time, it can also reflect the severity of the disease of the weak shrimps. The weak shrimps are usually piled together due to the decrease in vitality or avoidance, forming multi-layer stacking, which reflects the risk degree of cross-infection of the weak shrimp cluster, and reflects the deterioration degree of the local environment. Therefore, the application evaluates the health abnormality characteristic coefficient by the shortest horizontal distance of the smallest shrimp cluster from the obstacle and the corresponding stacking amount to represent the disease deterioration degree of the smallest shrimp cluster, which provides data support for subsequent determination of whether to clean the smallest shrimp cluster. The application improves the accuracy of health monitoring of the shrimp group on the basis of real-time and continuous monitoring of the shrimp group, and reduces the influence of shrimp clusters with possible health problems on the whole shrimp group through hierarchical management of the shrimp clusters, thereby ensuring the overall health level of the shrimp group.

[0019] Especially, based on locking the high-risk weak area, the application also considers the possible health risks of the remaining shrimp clusters, identifies the abnormal approaching behavior of the remaining shrimp clusters to the smallest shrimp cluster, analyzes and verifies the abnormal approaching behavior, and excludes random swimming interference through the vector direction corresponding to the swimming trajectory of the remaining shrimp clusters, to accurately identify the abnormal approaching behavior. Then, the application quantifies the initiative of the remaining shrimp clusters to approach the smallest shrimp cluster and the attraction degree of other factors such as stress secretions induced by weak shrimps or release of chemical signals by pathogens to other shrimp clusters through the distance shortening amount between the remaining shrimp clusters and the smallest shrimp cluster, so that the remaining shrimp clusters may have a cross-infection risk. Therefore, the application isolates the remaining shrimp clusters in time based on the above considerations. Thus, the application improves the accuracy of health monitoring of the shrimp group on the basis of real-time and continuous monitoring of the shrimp group, and reduces the influence of shrimp clusters with possible health problems on the whole shrimp group through hierarchical management of the shrimp clusters, thereby ensuring the overall health level of the shrimp group. BRIEF DESCRIPTION OF DRAWINGS

[0020] Figure 1 A step schematic diagram of the intelligent identification shrimp live health monitoring method based on a large model of the application embodiment; Figure 2 A logic determination diagram for marking the shrimp group of the application embodiment; Figure 3 A logic determination diagram for determining whether to clean the smallest shrimp cluster of the application embodiment; Figure 4 A logic determination diagram for determining whether to isolate the remaining shrimp clusters of the application embodiment. DETAILED DESCRIPTION

[0021] In order to make the objects, technical schemes and advantages of the present application clearer, the following further describes the present application with reference to the accompanying drawings; it should be understood that the specific embodiments described herein are only used to explain the present application and should not be used to limit the present application.

[0022] The preferred embodiments of the present application are described below with reference to the accompanying drawings. Those skilled in the art should understand that these embodiments are only used to explain the technical principles of the present application and are not used to limit the protection scope of the present application.

[0023] It should be noted that in the description of the present application, the terms indicating the direction or positional relationship of "inner" and the like are based on the direction or positional relationship shown in the drawings, which is only for the convenience of description, and does not indicate or imply that the device or element must have a specific orientation, be constructed and operated in a specific orientation, and therefore cannot be understood as a limitation on the present application.

[0024] In addition, it should also be noted that in the description of the present application, unless otherwise explicitly specified and limited, the term "mounting" should be understood broadly. Those skilled in the art can understand the specific meaning of the above-mentioned term in the present application according to the specific circumstances.

[0025] Please refer to Figure 1 Fig. 1 shows a step schematic diagram of a shrimp group living body health monitoring method based on a large model according to an embodiment of the present application, and the shrimp group living body health monitoring method based on a large model according to the embodiment of the present application comprises: Step S1, continuously acquiring shrimp group images in a culture pond to extract active distribution features of the shrimp group, wherein the active distribution features include a moving offset shrimp cluster density and a moving depth change amount of a corresponding shrimp cluster; Step S2, analyzing active offset representation values of the shrimp group according to the active distribution features combined with a moving speed of the shrimp cluster to label the shrimp group; Step S3, in response to the labeling result, analyzing the shrimp group, including, identifying a trajectory moving discrete stage of the shrimp group, analyzing a plurality of shrimp clusters in the trajectory moving discrete stage, including, acquiring each branch trajectory corresponding to the trajectory moving discrete stage, extracting moving data of a corresponding minimum shrimp cluster in each branch trajectory, calling a horizontal shortest distance of the corresponding minimum shrimp cluster from an obstacle in the culture pond and a stacking amount of the minimum shrimp cluster to evaluate a health abnormality representation coefficient of the minimum shrimp cluster to determine whether to clean the minimum shrimp cluster; acquiring moving trajectories of a plurality of remaining shrimp clusters, determining trajectory trend values of each of the remaining shrimp clusters, and matching with a trajectory trend reference range to determine whether to isolate the remaining shrimp clusters.

[0026] Specifically, the way to obtain the image of the shrimp group in the culture pond is not specifically limited, as long as the image of the shrimp group can be collected, a number of rotating cameras can be installed on the inner wall of the culture pond, the camera angle can be adjusted through the holder to cover a larger collection range, and a pollution-proof cover can be used to prevent algae from adhering to the camera and affecting the collection of the shrimp group image. Of course, other ways to obtain the image of the shrimp group can also be used. Among them, the spatial-temporal features of the shrimp group, such as the density of the moving offset shrimp cluster, the stacking amount, etc., can be learned from the image through the model of ViT (Vision Transformer).

[0027] Specifically, the process of analyzing the active offset representation value of the shrimp group according to the active distribution feature and the swimming speed of the shrimp cluster includes, The sum of the ratio of the density of the moving offset shrimp cluster to the shrimp cluster density threshold and the ratio of the swimming depth change amount of the corresponding shrimp cluster to the swimming depth change amount threshold is taken as the first active offset feature. The ratio of the swimming speed threshold to the swimming speed of the shrimp cluster is taken as the second active offset feature. The first active offset feature and the second active offset feature are weighted and summed to determine the active offset representation value.

[0028] Specifically, in the actual swimming process of the shrimp group, the number of moving offset shrimp clusters in the shrimp group can intuitively reflect the degree of health risk of the whole shrimp group, and the depth change of the corresponding swimming position of the moving offset shrimp cluster can clearly reflect the severity of the health damage of the corresponding shrimp cluster. Therefore, in the implementation, the active distribution feature, i.e., the density of the moving offset shrimp cluster and the swimming depth change amount of the corresponding shrimp cluster, is given priority, so the first active offset feature calculated based on the active distribution feature is given a slightly higher weight. Therefore, when weighted and summed, the weight of the first active offset feature is set to 0.6, and the weight of the second active offset feature is set to 0.4. In the embodiment, the purposes of setting the shrimp cluster density threshold value, the swimming depth change amount threshold value and the swimming speed threshold value are all to represent the severity of the dispersion of the swimming trajectory of the shrimp group, and then represent the severity of the health damage of the weak shrimp cluster. The historical data of the shrimp cluster density, the swimming depth change amount and the swimming speed of the shrimp group are extracted by calling the historical shrimp group images in the aquaculture ponds, the mean values of the shrimp cluster density, the swimming depth change amount and the swimming speed are solved, the shrimp cluster density threshold value is determined as the product of the mean value of the shrimp cluster density and a density deviation coefficient, the swimming depth change amount threshold value is determined as the product of the mean value of the swimming depth change amount and a depth deviation coefficient, and the swimming speed threshold value is determined as the product of the mean value of the swimming speed and a speed deviation coefficient, wherein the density deviation coefficient is selected in the interval [1.15, 1.2], the depth deviation coefficient is selected in the interval [1.2, 1.25], and the speed deviation coefficient is selected in the interval [0.9, 0.95].

[0029] Specifically, referring to FIG. 1, Figure 2 As shown in FIG. 1, the logic judgment diagram for marking the shrimp group by the embodiment of the present application comprises, If the active deviation representation value of the shrimp group is greater than or equal to the active deviation representation threshold value, the shrimp group is set with a high deviation degree label. If the active deviation representation value of the shrimp group is less than the active deviation representation threshold value, the shrimp group does not need to be marked.

[0030] The active deviation representation threshold value is selected in the interval [1.69, 1.74] Specifically, based on the factors such as the decline of the physiological function and the decrease of the environmental adaptation ability of the weak shrimp with health problems, the weak shrimp separates from the shrimp group and forms a separate swimming trajectory, the swimming active distribution of the shrimp group in the aquaculture pond is considered, the shrimp cluster density of the swimming deviation of the shrimp group is separated to reflect the damage degree of the weak shrimp to the whole shrimp group. Compared with the normal shrimp group uniformly distributed in the upper layer or the flow area of the water body, the weak shrimp usually gathers in the deep places such as the bottom and the corner of the aquaculture pond, so the swimming depth change amount of the shrimp cluster can reflect the degree of physical exhaustion and the degree of decline of the swimming ability of the weak shrimp, and the swimming speed can directly reflect the activity level of the shrimp cluster. Therefore, the present application can comprehensively capture the activity and health status of the whole shrimp group by combining the above-mentioned features, represent the severity of the dispersion of the swimming trajectory of the shrimp group according to the active offset characteristic value, and the offset degree of the swimming trajectory of the dispersed shrimp cluster, provide data support for subsequent marking of the shrimp group, and then adaptively analyze the shrimp group. The present application improves the accuracy of the health monitoring of the shrimp group on the basis of real-time and continuous monitoring of the shrimp group, and reduces the influence of the shrimp cluster with possible health problems on the whole shrimp group through hierarchical management of the shrimp cluster, thereby ensuring the overall health level of the shrimp group.

[0031] Specifically, in response to the marking result, the shrimp group is analyzed, including, If any shrimp group is set with a high offset degree label, the shrimp group is analyzed.

[0032] Specifically, the process of identifying the trajectory swimming dispersion stage of the shrimp group includes, Calling a shrimp group image to identify the swimming trajectory of the shrimp group; If the swimming trajectory meets the trajectory dispersion condition, it is determined that the shrimp group enters the trajectory swimming dispersion stage; Wherein, the trajectory meets the trajectory forming condition includes that two or more swimming trajectories appear and are maintained for a predetermined time length.

[0033] It can be understood that due to the influence of shrimp species and environmental factors, the swimming speed of shrimps is different. For example, the normal swimming speed of South American white shrimp is 5cm / s-15cm / s, and when the density of the shrimp group in the breeding environment is >300 tails / m³, the swimming speed of South American white shrimp will decrease to 3cm / s-8cm / s due to limited activity. Therefore, in order to intuitively observe the swimming offset of the shrimp group, the predetermined time length maintained by the swimming trajectory is selected within the interval [30s, 1min].

[0034] Specifically, the process of evaluating the health abnormality characteristic coefficient of the minimum shrimp cluster includes, The ratio of the horizontal shortest distance threshold value to the horizontal shortest distance of the minimum shrimp cluster from the obstacle in the breeding pond is taken as the first health abnormality feature; The ratio of the stacking amount of the minimum shrimp cluster to the stacking amount threshold value is taken as the second health abnormality feature; The sum of the first health abnormality feature and the second health abnormality feature is taken as the health abnormality characteristic coefficient.

[0035] In this embodiment, the purpose of setting the horizontal shortest distance threshold and the stacking amount threshold is to represent the severity of the disease deterioration of the weak shrimp. By obtaining a plurality of historical shrimp group images in the breeding pond corresponding to the disease, the historical data of the horizontal shortest distance of the minimum shrimp cluster from the obstacle in the breeding pond and the historical data of the stacking amount of the minimum shrimp cluster are extracted, the mean value of the horizontal shortest distance and the mean value of the stacking amount are solved, and the purpose of setting the two thresholds is to determine the horizontal shortest distance threshold as the product of the mean value of the horizontal shortest distance and the distance deviation coefficient, and determine the stacking amount threshold as the product of the mean value of the stacking amount and the stacking deviation coefficient, wherein the distance deviation coefficient is selected in the interval [0.9, 0.95], and the stacking deviation coefficient is selected in the interval [1.1, 1.15]. Wherein, the stacking height of the minimum shrimp cluster is taken as the stacking amount.

[0036] Specifically, the present application differentiates the processing and monitoring of the shrimp group with a high deviation degree label, focuses on analyzing the discrete swimming trajectory of the whole shrimp group, locks the high-risk weak area, realizes the dynamic positioning of the weak shrimp cluster and prioritizes the waste disposal and other processing. The weak shrimps are easily pushed to the vicinity of the obstacles, such as the corner of the pond, the sewage outlet, etc., due to their inability to resist the water flow impact caused by their physical exhaustion or nerve damage, which shortens the horizontal distance between them and the obstacles in the breeding pond, and also reflects the severity of the disease of the weak shrimps. The weak shrimps are usually stacked together due to the decline in vitality or avoidance, forming multi-layer stacking, which reflects the risk degree of cross-infection of the weak shrimp cluster and the deterioration degree of the local environment. Therefore, the present application evaluates the health abnormality representation coefficient by the horizontal shortest distance of the minimum shrimp cluster from the obstacle and the corresponding stacking amount to represent the disease deterioration degree of the minimum shrimp cluster, which provides data support for subsequent determination of whether to dispose the minimum shrimp cluster. The present application improves the accuracy of health monitoring of the shrimp group on the basis of real-time and continuous monitoring of the shrimp group, and reduces the influence of the shrimp cluster with possible health problems on the whole shrimp group through hierarchical management of the shrimp cluster, thereby ensuring the health level of the whole shrimp group.

[0037] Specifically, please refer to Figure 3 As shown in the figure, it is a logic determination diagram for determining whether to dispose the minimum shrimp cluster according to the present application, and the determination whether to dispose the minimum shrimp cluster includes, If the health abnormality representation coefficient of the minimum shrimp cluster is greater than or equal to the health abnormality representation coefficient threshold, it is determined to dispose the minimum shrimp cluster. If the health abnormality representation coefficient of the minimum shrimp cluster is less than the health abnormality representation coefficient threshold, it is determined not to dispose the minimum shrimp cluster.

[0038] The health abnormality representation coefficient threshold is selected in the interval [2.18, 2.24].

[0039] Specifically, the process of determining the trajectory tendency value of each of the remaining shrimp clusters comprises, generating a plurality of vectors corresponding to the swimming trajectories of the plurality of remaining shrimp clusters, respectively; selecting a plurality of remaining target shrimp clusters whose vector directions of each of the vectors remain unchanged within a predetermined time; determining the distance reduction amount of each of the remaining target shrimp clusters from the minimum shrimp cluster within the predetermined time; taking the distance reduction amount as the trajectory tendency value.

[0040] It can be understood that the new sick shrimp with diseases has a short infection time, a slight disease state, and an activity ability that has not decreased significantly. The old sick shrimp with diseases will avoid the old sick shrimp due to sensory signals, and then swim in other directions relative to the direction in which the old sick shrimp is located. However, as the infection time increases, the disease state becomes severe, and at the same time, the environment stress of high-density farming or a farming pond will cause the new sick shrimp to be passively gathered in the same area as the old sick shrimp. Therefore, in the embodiment, the distance reduction amount between the remaining shrimp clusters maintaining single-direction swimming and the minimum shrimp cluster is considered to reflect the trajectory convergence degree of the new sick shrimp and the old sick shrimp, and then the disease deterioration degree of the new sick shrimp is evaluated.

[0041] In the embodiment, the distance reduction amount is taken as the difference between the distance of the remaining target shrimp cluster from the minimum shrimp cluster at the start time of the predetermined time and the distance of the remaining target shrimp cluster from the minimum shrimp cluster at the end time of the predetermined time. Since the swimming speed of the sick shrimp with diseases will decrease accordingly, in order for the collected data to be characteristic, the predetermined time is selected within the interval [7 min, 10 min].

[0042] Specifically, referring to Figure 4 which is a logic determination diagram for determining whether to isolate the remaining shrimp clusters according to the embodiment of the application. The determination of whether to isolate the remaining shrimp clusters comprises, if the trajectory tendency value of any of the remaining target shrimp clusters is not in the trajectory tendency reference range, it is determined that the remaining target shrimp cluster is isolated; if the trajectory tendency value of any of the remaining target shrimp clusters is in the trajectory tendency reference range, it is determined that the remaining target shrimp cluster does not need to be isolated; The trajectory tendency reference range is pre-set.

[0043] In this embodiment, by acquiring a plurality of historical data of the related shrimp cluster with disease, the distance shortening amount historical data of a plurality of remaining target shrimp clusters from the minimum shrimp cluster is determined, the maximum value and the minimum value of the distance shortening amount in the distance shortening amount historical data are determined, the maximum value is taken as the upper limit of the interval of the trajectory trend reference range, and the minimum value is taken as the lower limit of the interval of the trajectory trend reference range, wherein the trajectory trend reference range is a closed area.

[0044] Specifically, on the basis of locking the high-risk weak area, the present application also considers the health risks that the remaining shrimp clusters may have, identifies the abnormal approaching behavior of the remaining shrimp clusters to the minimum shrimp cluster, analyzes and verifies the abnormal approaching behavior, excludes random wandering interference through the vector direction corresponding to the moving trajectory of the remaining shrimp cluster, accurately identifies the abnormal approaching behavior, and further quantifies the distance shortening amount between the remaining shrimp cluster and the minimum shrimp cluster to reflect the initiative of the remaining shrimp cluster to approach the minimum shrimp cluster, that is, the weak shrimp cluster, and the degree of attraction of other factors such as stress exudate induction or pathogen release of chemical signals to other shrimp clusters, so that the remaining shrimp cluster may have a cross-infection risk. Therefore, the present application isolates the remaining shrimp cluster in time based on the above considerations, and further improves the accuracy of the health monitoring of the shrimp cluster based on the real-time and continuous monitoring of the shrimp cluster, reduces the influence of the shrimp cluster that may have health problems on the whole shrimp cluster, and ensures the health level of the whole shrimp cluster.

[0045] Specifically, it also includes setting a response prompt signal, In response to the minimum shrimp cluster being cleared or the remaining target shrimp cluster being isolated, a prompt signal is sent.

[0046] If the intelligent identification shrimp live health monitoring method based on a large model of the present application is realized in the form of a software function unit and sold or used as an independent product, it can be stored in a computer readable storage medium. Based on this understanding, the technical solution of the present application or the part that essentially contributes to the prior art or the whole or part of the technical solution can be embodied in the form of a software product, which is stored in a storage medium and includes a plurality of instructions for causing a computer device (which can be a personal computer, a server, or a network device) to execute all or part of the steps of the method described in various embodiments of the present application. The aforementioned storage medium includes: a U disk, a mobile hard disk, a read-only memory (ROM, Read-Only Memory), a random access memory (RAM, Random Access Memory), a magnetic disk or an optical disk, and various program code storage media.

[0047] The technical scheme of the present application has been described in combination with the preferred embodiments shown in the drawings, but it is easy for those skilled in the art to understand that the protection scope of the present application is obviously not limited to these specific embodiments. Those skilled in the art can make equivalent changes or replacements to the related technical features without departing from the principles of the present application, and the technical schemes after the changes or replacements will all fall within the protection scope of the present application.

Claims

1. A method for intelligently identifying the health of live shrimps based on a large model, characterized in that: include: Continuously acquiring images of shrimp swarms in the aquaculture pond to extract active distribution characteristics of the shrimp swarms, the active distribution characteristics including the density of shrimp clusters with swimming offsets and the change in swimming depth of the corresponding shrimp clusters; Analyzing the activity deviation characterization value of the shrimp group according to the activity distribution characteristics and the swimming speed of the shrimp cluster to mark the shrimp group; In response to the marking results, the shrimp population is analyzed, include, Identifying discrete stages of the shrimp swarm's trajectory swimming, and analyzing a plurality of shrimp clusters in the discrete stages of the trajectory swimming, including obtaining each branch trajectory corresponding to the discrete stages of the trajectory swimming, extracting swimming data of the smallest shrimp cluster corresponding to each branch trajectory, and calling the shortest horizontal distance between the corresponding smallest shrimp cluster and an obstacle in the breeding pond and the stacking amount of the smallest shrimp cluster to evaluate the health abnormality characterization coefficient of the smallest shrimp cluster, so as to determine whether to clear the smallest shrimp cluster; The swimming trajectories of several remaining shrimp clusters are obtained, and the trajectory trend value of each of the remaining shrimp clusters is determined and matched with the trajectory trend reference range to determine whether to isolate the remaining shrimp clusters.

2. The method for intelligently identifying the health of living shrimps based on a large model according to claim 1 is characterized in that: The process of analyzing the active offset characterization value of the shrimp group based on the active distribution characteristics and the swimming speed of the shrimp cluster includes: The sum of the ratio of the shrimp cluster density of the swimming deviation to the shrimp cluster density threshold and the ratio of the swimming depth change of the corresponding shrimp cluster to the swimming depth change threshold is used as the first active deviation feature; The ratio of the swimming speed threshold to the swimming speed of the shrimp cluster is used as the second active offset feature; The first active offset feature and the second active offset feature are weightedly summed to determine the active offset representation value.

3. The method for intelligently identifying the health of living shrimps based on a large model according to claim 2 is characterized in that: Marking the shrimp population, comprising: If the active excursion characterization value of the shrimp group is greater than or equal to the active excursion characterization threshold, a high excursion degree label is set for the shrimp group.

4. The method for intelligently identifying the health of living shrimps based on a large model according to claim 3 is characterized in that: In response to the labeling results, the shrimp population is analyzed, including: If there is any shrimp group that is labeled with a high degree of deviation, the shrimp group is analyzed.

5. The method for intelligently identifying the health of living shrimps based on a large model according to claim 1 is characterized in that: The process of identifying discrete stages of the shrimp's swimming trajectory, include, Retrieving images of a swarm of shrimps to identify the swimming trajectories of the swarm of shrimps; If the swimming trajectory meets the trajectory discrete condition, it is determined that the shrimp group enters the trajectory discrete stage; The trajectory meeting the trajectory forming condition includes two or more swimming trajectories appearing and maintaining a predetermined time period.

6. The method for intelligently identifying the health of living shrimps based on a large model according to claim 1, characterized in that: The process of evaluating the health abnormality characterization coefficient of the minimum shrimp cluster comprises: The ratio of the horizontal shortest distance threshold to the horizontal shortest distance between the smallest shrimp cluster and the obstacle in the culture pond is used as the first health abnormality feature; The ratio of the stacking volume of the smallest shrimp cluster to the stacking volume threshold is used as the second health abnormality feature; The sum of the first health abnormality feature and the second health abnormality feature is used as the health abnormality characterization coefficient.

7. The method for intelligently identifying the health of living shrimps based on a large model according to claim 6, characterized in that: Determine whether to clear the smallest shrimp cluster, include, If the health abnormality characterization coefficient of the smallest shrimp cluster is greater than or equal to the health abnormality characterization coefficient threshold, it is determined that the smallest shrimp cluster is to be cleared.

8. The method for intelligently identifying the health of living shrimps based on a large model according to claim 1, characterized in that: The process of determining the trajectory trend value of each of the remaining shrimp clusters includes: Generate several vectors corresponding to the swimming trajectories of the remaining shrimp clusters respectively; Selecting a number of other target shrimp clusters corresponding to the vector directions of the vectors remaining unchanged within a predetermined time; determining an amount of shortening of the distance between each of the remaining target shrimp clusters and the smallest shrimp cluster within the predetermined time; The shortened distance is used as the trajectory trend value.

9. The method for intelligently identifying the health of living shrimps based on a large model according to claim 8, characterized in that: Determining whether to isolate the remaining shrimp clusters includes: If the trajectory trend value of any other target shrimp cluster is not within the trajectory trend reference range, it is determined that the other target shrimp cluster is to be isolated; Wherein, the trajectory trend reference range is pre-set.

10. The method for intelligently identifying the health of living shrimps based on a large model according to claim 1, characterized in that: Also included is setting a response prompt signal, In response to the smallest shrimp cluster being eliminated or the remaining target shrimp clusters being isolated, a prompt signal is issued.

Images