A lightweight fish identification method and system based on edge intelligence

Abstract

The present application relates to the field of fishery management technology, and in particular to a lightweight fish identification method and system based on edge intelligence, which collects information of fish samples to be identified; determines that there is plan information associated with the requesting person's attributes, and then collects the plan time and the second category ID included in the plan information; collects the daily shooting duration associated with each activity video, and determines the ratio between the number of associated valid items and the daily shooting duration as the effective duration associated with each activity video in multiple activity videos; collects the shooting cycle associated with each activity video; collects the number of IDs of activity videos whose effective duration is greater than the associated shooting cycle in multiple activity videos; if the number of IDs is greater than zero, the fish sample to be identified is determined as a risk target fish. The present invention can effectively target the safety risks of farmed fish in aquaculture waters caused by the mixing of other fish species in fish schools, and automatically distinguish the fish in the waters to better raise fish.

Description

Technical Field

[0001] The present application relates to the field of fishery management technology, and in particular to a lightweight fish identification method and system based on edge intelligence. Background Art

[0002] Aquaculture is an important part of my country's fishery and a major growth point for fishery development. However, aquaculture usually requires several breeding ponds to feed different types of fish separately, and farmers use artificial feeding to feed the fish, which is time-consuming and laborious.

[0003] At the same time, in current aquaculture, especially in open waters, other fish species are often mixed in with the fish schools, posing safety risks to farmed fish in the aquaculture waters. How to automatically distinguish the fish in the waters to better raise the fish is an urgent problem to be solved in the current aquaculture industry. Summary of the invention

[0004] To achieve the above objectives, this application provides the following technical solutions:

[0005] According to a first aspect of the present invention, the present invention claims a lightweight fish identification method based on edge intelligence, comprising:

[0006] Collecting fish samples to be identified, the fish samples to be identified include multiple activity videos, the scale of fish swimming associated with each activity video in the multiple activity videos and the magnitude of the fish swimming scale, the attributes of the requesting person, the requesting time, and the first category ID;

[0007] Determine whether there is plan information associated with the requesting person attribute in the second plan log set;

[0008] If the second plan log set contains the plan information associated with the requester attribute, the plan time and the second category ID included in the plan information are collected; the plan time is the time of the future request determined based on the fish indicated by the requester attribute during the historical request;

[0009] If the request time is earlier than the planned time, and the first category ID is consistent with the second category ID, then based on the fish swimming scale associated with each of the multiple activity videos, the valid number of items associated with each of the multiple activity videos is obtained, where the valid number of items associated with each of the multiple activity videos is obtained by tracking and detecting the fish magnitude of the fish swimming scale associated with the activity video;

[0010] Collecting the daily shooting duration associated with each of the multiple activity videos in the first plan log set;

[0011] Determine the ratio between the number of valid items associated with each of the multiple activity videos and the daily shooting duration as the valid duration associated with each of the multiple activity videos;

[0012] Collecting the shooting period associated with each of the multiple activity videos; the shooting period associated with each activity video refers to the period during which the fish indicated by the activity video can be active at most;

[0013] Collecting the number of IDs of the multiple activity videos whose effective duration is greater than the associated shooting period;

[0014] If the number of IDs is greater than zero, the fish sample to be identified is determined as a risky target fish.

[0015] Furthermore, the collecting of fish samples to be identified includes:

[0016] Acquire a judgment operation from a mobile device, wherein the judgment operation includes the fish sample to be identified, and the judgment operation is obtained when the mobile device judges a trigger operation of a judgment control on a fish swimming environment;

[0017] After determining the fish sample to be identified as a risk target fish, the method further includes:

[0018] Determine a risk activity video whose effective duration associated with each activity video in the multiple activity videos is greater than a shooting period;

[0019] An early warning notification is transmitted to the mobile device, wherein the early warning notification includes the risk activity video, and the early warning notification is used to warn the mobile device to modify the fish swimming scale associated with the risk activity video in the fish sample to be identified and / or the order of fish in the fish swimming scale.

[0020] Further, the fish sample to be identified also includes body size, the first planned log set includes multiple log records, and the daily shooting duration associated with each of the multiple activity videos collected in the first planned log set includes:

[0021] Determining a target log record associated with the body size from among the plurality of log records, wherein different log records among the plurality of log records store different daily shooting durations associated with each of the plurality of activity videos;

[0022] The daily shooting duration associated with each of the multiple activity videos is collected in the target log record.

[0023] Further, the fish sample to be identified also includes a target fish ID. After collecting the number of IDs of the active videos whose effective duration is greater than the associated shooting cycle in the multiple active videos, the method further includes:

[0024] If the number of IDs is zero, the fish sample to be identified is determined as a valid target fish;

[0025] The valid target fish is stored in the third planned log set so that when a feeding request including a target fish ID is obtained, if the target fish ID of a target valid target fish in the third planned log set is consistent with the target fish ID in the feeding request, the feeding request is fed back.

[0026] Further, based on the fish swimming scale associated with each of the multiple activity videos, obtaining the valid number of items associated with each of the multiple activity videos includes:

[0027] Based on the fish magnitude of the fish swimming scale associated with each of the plurality of activity videos, collecting a mapping value associated with the fish magnitude from a fourth planned log set;

[0028] The fish swimming scale associated with each of the multiple activity videos and the mapping value associated with the fish magnitude of the fish swimming scale are multiplied to obtain the valid number of items associated with each of the multiple activity videos.

[0029] According to the second aspect of the present invention, the present invention seeks protection for a lightweight fish identification system based on edge intelligence, characterized in that the system includes a module for executing the lightweight fish identification method based on edge intelligence.

[0030] The present application relates to the field of fishery management technology, and in particular to a lightweight fish identification method and system based on edge intelligence, which collects information of fish samples to be identified; determines that there is plan information associated with the requesting person's attributes, and then collects the plan time and the second category ID included in the plan information; collects the daily shooting duration associated with each activity video, and determines the ratio between the number of associated valid items and the daily shooting duration as the effective duration associated with each activity video in multiple activity videos; collects the shooting cycle associated with each activity video; collects the number of IDs of activity videos whose effective duration is greater than the associated shooting cycle in multiple activity videos; if the number of IDs is greater than zero, the fish sample to be identified is determined as a risk target fish. The present invention can effectively target the safety risks of farmed fish in aquaculture waters caused by the mixing of other fish species in fish schools, and automatically distinguish the fish in the waters to better raise fish. BRIEF DESCRIPTION OF THE DRAWINGS

[0031] Figure 1 A schematic diagram of a flow chart of a lightweight fish identification method based on edge intelligence as claimed in an embodiment of the present application;

[0032] Figure 2A second flow chart of a lightweight fish identification method based on edge intelligence claimed in an embodiment of the present application;

[0033] Figure 3 A structural module diagram of a lightweight fish identification system based on edge intelligence as claimed in an embodiment of the present application. DETAILED DESCRIPTION

[0034] The following will be combined with the drawings in the embodiments of the present application to clearly and completely describe the technical solutions in the embodiments of the present application. Obviously, the described embodiments are only part of the embodiments of the present application, not all of the embodiments. Based on the embodiments in the present application, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of this application.

[0035] The terms "first", "second" and "third" in this application are only used for descriptive purposes, and cannot be understood as indicating or implying relative importance or implicitly indicating the number of indicated technical features. Thus, the features defined as "first", "second" and "third" can explicitly or implicitly include at least one of the features. In the description of this application, the meaning of "multiple" is at least two, such as two, three, etc., unless otherwise clearly and specifically defined. All directional indications in the embodiments of the present application (such as up, down, left, right, front, back...) are only used to explain the relative position relationship, movement, etc. between the components under a certain specific posture (as shown in the accompanying drawings). If the specific posture changes, the directional indication also changes accordingly. In addition, the terms "including" and "having" and any of their variations are intended to cover non-exclusive inclusions. For example, a process, method, system, product or device that includes a series of steps or modules is not limited to the listed steps or modules, but optionally also includes steps or modules that are not listed, or optionally also includes other steps or modules inherent to these processes, methods, products or devices.

[0036] Reference to "embodiments" herein means that a particular feature, structure, or characteristic described in conjunction with the embodiments may be included in at least one embodiment of the present application. The appearance of the phrase in various locations in the specification does not necessarily refer to a consistent embodiment, nor is it an independent or alternative embodiment that is mutually exclusive with other embodiments. It is explicitly and implicitly understood by those skilled in the art that the embodiments described herein may be combined with other embodiments.

[0037] See also Figure 1 , Figure 1 FIG. 1 is a flow chart of a lightweight fish identification method based on edge intelligence provided by an embodiment of the present invention. Specifically, Figure 1As shown, the lightweight fish identification method based on edge intelligence in the embodiment of the present invention may include but is not limited to the following steps:

[0038] S101. The server collects fish samples to be identified, where the fish samples to be identified include multiple activity videos, the swimming scale of fish associated with each activity video in the multiple activity videos, and the fish order of magnitude of the swimming scale of the fish.

[0039] Among them, the fish samples to be identified refer to the aquatic documents provided by practicing breeders and practicing assistant breeders (referred to as breeders) to breeders during diagnosis and treatment activities, which are inspected by aquatic product professionals who have obtained the qualifications for aquatic product professional positions, and serve as breeding certificates for breeders. Nutritionists can prepare feed for breeders based on the fish samples to be identified. Correspondingly, breeders can obtain fish associated with the fish samples to be identified from the feed warehouse and eat them according to the amount based on the fish samples to be identified. In one implementation, the fish samples to be identified can be target fish that have been fed that day, or target fish that have not been fed that day.

[0040] By judging whether the target fish that has not been fed that day is a risky target fish, if the target fish is determined to be a risky target fish before the keeper feeds it, new valid target fish can be provided to the keeper to avoid adverse effects on the keeper's health due to the active risky target fish.

[0041] Among them, risk target fish refers to fish that do not meet the requirements, such as wild fish, unhealthy fish, etc.;

[0042] The activity video is used to uniquely ID the fish indicated in the fish sample to be identified. For example, the activity video can be the name of the fish or the number of the fish, etc. The scale of fish swimming associated with each activity video is the total number of fish indicated by the activity video that the salesperson gave to the breeder at the time of this request, and the fish order of magnitude of the fish swimming scale is the order of magnitude of the total number. In one implementation, each type of fish can have a variety of different orders of magnitude in different target fish. For example, the content of the fish sample to be identified can be as follows: fish a: 1 group, fish b: 2 groups, fish c: 3 groups. At this time, the fish order of magnitude of fish a is 102, the fish order of magnitude of fish b is 103, and the fish order of magnitude of fish c is 104.

[0043] In one implementation, the specific implementation method of the server collecting the fish sample to be identified can be: obtaining a judgment operation from a mobile device, the judgment operation includes the fish sample to be identified, and the judgment operation is obtained when the mobile device judges the trigger operation of the judgment control on the fish swimming environment. In this way, after the salesperson determines the fish in the pre-sale fish pond, but before transmitting the target fish to the feed warehouse system, he can click the judgment control to start the automatic judgment program for the target fish, which is conducive to improving the safety of breeding.

[0044] In one implementation, the judgment operation can be automatically obtained after the mobile device judges that the salesperson has completed the input in the target fish input environment. In this way, on the one hand, the automatic judgment program for the target fish can be started without manually clicking the judgment control, which is conducive to improving the judgment efficiency; on the other hand, it can also avoid the situation where the salesperson forgets to click the judgment control and cannot judge the target fish, which is conducive to improving the breeding safety.

[0045] In one implementation, the fish sample to be identified may also include the attributes of the requesting person. After collecting the fish sample to be identified, the server may also perform the following steps: the server collects invalid activity videos associated with the attributes of the requesting person of the fish sample to be identified that are pre-stored, and determines whether there are invalid activity videos in the multiple activity videos of the fish sample to be identified. If there are no invalid activity videos in the multiple activity videos of the fish sample to be identified, then a step is triggered to obtain the number of valid items associated with each of the multiple activity videos based on the fish swimming scale associated with each of the multiple activity videos.

[0046] In one implementation, when the fish indicated by the requesting person's attributes activates the fish indicated by the invalid activity video associated with the requesting person's attributes of the fish sample to be identified, an invalid reaction will be generated. Among them, the invalid activity video associated with the requesting person's attributes of the fish sample to be identified can be obtained based on the historical shooting records associated with the requesting person's attributes. In one implementation, different requesting person attributes can be associated with different invalid activity videos, or consistent invalid activity videos, which is not limited by the embodiment of the present invention. In one implementation, the fish indicated by the invalid activity video can include, but are not limited to: fish with low activity, and fish inconsistent with group activities.

[0047] In one implementation, if there is an invalid activity video among the multiple activity videos of the fish sample to be identified, the judgment program is exited, or a prompt message is output, and the prompt message is used to prompt that there is an invalid activity video in the fish sample to be identified, so that the salesperson can re-provide a new target fish that does not include an invalid activity video for the fish indicated by the requester attribute, which is beneficial to improving the breeding safety. In one implementation, the requester attribute can be associated with one or more invalid activity videos. When the fish ID is associated with multiple invalid activity videos, if there is at least one invalid activity video among the multiple activity videos of the fish sample to be identified, the judgment program is exited, or a prompt message is output.

[0048] The embodiment of the present invention triggers the step of obtaining the number of valid items associated with each of the multiple activity videos based on the fish swimming scale associated with each of the multiple activity videos only when there are no invalid activity videos in the multiple activity videos of the fish sample to be identified, that is, continues to execute the judgment procedure; and when there are invalid activity videos in the multiple activity videos of the fish sample to be identified, the judgment procedure is terminated in advance, which can improve the judgment efficiency and help improve the breeding safety.

[0049] S102. The server obtains the valid number of items associated with each of the multiple activity videos based on the fish swimming scale associated with each of the multiple activity videos. The valid number of items associated with each activity video is obtained after tracking and detecting the fish magnitude of the fish swimming scale associated with the activity video.

[0050] Specifically, the specific implementation method of the server obtaining the valid number of items associated with each of the multiple activity videos based on the fish swimming scale associated with each of the multiple activity videos can be: based on the order of fish of the fish swimming scale associated with each of the multiple activity videos, the server collects the mapping value associated with the order of fish from the fourth plan log set, multiplies the fish swimming scale associated with each of the multiple activity videos and the mapping value associated with the order of fish of the fish swimming scale, to obtain the valid number of items associated with each of the multiple activity videos.

[0051] As can be seen from the aforementioned step S101, in one implementation, the swimming scale of each fish can have multiple different orders of magnitude in different target fish. Through the mapping value associated with the fish order of magnitude of the swimming scale of the fish indicated by each activity video, the different fish orders of magnitude of the swimming scale of the fish with the same activity video in different target fish can be converted into a unified order of magnitude. In this way, it is easy to judge whether the fish indicated by the activity video is excessive, which is conducive to improving the judgment efficiency of risk target fish.

[0052] In one implementation, the server may collect fish specifications and / or water specifications associated with each of multiple activity videos of the fish sample to be identified, and based on the fish specifications and / or water specifications associated with each of the multiple activity videos of the fish sample to be identified, and the order of magnitude of fish swimming scale associated with each of the multiple activity videos of the fish sample to be identified, obtain a mapping value associated with the order of magnitude of fish, and store the mapping value associated with the order of magnitude of fish in a fourth planned log set.

[0053] In one implementation, the server obtains the mapping value associated with the fish order of magnitude based on the fish specifications and / or water specifications associated with each of the multiple activity videos of the fish sample to be identified, and the fish order of magnitude of the fish swimming scale associated with each of the multiple activity videos of the fish sample to be identified. The specific implementation method can be: the server collects the preset order of magnitude associated with each of the multiple activity videos of the fish sample to be identified, and based on the fish specifications and / or water specifications associated with each of the multiple activity videos of the fish sample to be identified, converts the fish order of magnitude of the fish swimming scale associated with each of the multiple activity videos of the fish sample to be identified into an associated preset order of magnitude, thereby obtaining the mapping value associated with the fish order of magnitude.

[0054] In one implementation, the preset magnitudes associated with different activity videos in the fish samples to be identified may be different. In one implementation, the preset magnitudes associated with consistent activity videos in different target fish are consistent, and the consistent activity videos in different target fish may be associated with different magnitudes of fish. When the magnitudes of fish associated with the consistent activity videos in different target fish are different, the associated mapping values ​​are also different.

[0055] S103: The server collects the daily shooting duration associated with each of the multiple activity videos in the first plan log set.

[0056] Among them, the daily shooting duration associated with each activity video is the number of fish indicated by the activity video that cannot be exceeded when active in one day. If the daily effective number of fish associated with the activity video is greater than the daily shooting duration, it will cause overfeeding when the breeders feed according to the daily effective number of fish, which will have an adverse effect on the health of the breeders.

[0057] In one implementation, the fish sample to be identified may also include body size and / or target gender, the first planned log set may include multiple log records, and the specific implementation method of the server collecting the daily shooting duration associated with each of the multiple activity videos in the first planned log set may be: determine the target log record associated with body size and / or target gender in the multiple log records of the first planned log set, different log records in the multiple log records store different daily shooting durations associated with each of the multiple activity videos, and collect the daily shooting duration associated with each of the multiple activity videos in the target log record. That is, based on the age and / or gender of the breeder, the daily shooting duration associated with the same activity video may be different.

[0058] S104: The server determines a ratio between the number of valid items associated with each of the multiple activity videos and the daily shooting duration as the valid duration associated with each of the multiple activity videos.

[0059] The effective duration associated with each activity video refers to the minimum period required for the fish to complete the activity video indicating the number of fish swimming.

[0060] S105: The server collects a shooting period associated with each of the multiple activity videos.

[0061] The shooting period associated with each activity video refers to the maximum period during which the fish indicated by the activity video can be active. If the active period exceeds the shooting period, it will have an adverse effect on the health of the breeder. In one implementation, the shooting period associated with the same activity video can be different based on the age and / or gender of the breeder.

[0062] S106. The server collects the number of IDs of the activity videos whose effective duration is greater than the associated shooting period from the multiple activity videos.

[0063] Specifically, the server can count the number of IDs of the activity videos whose effective duration is greater than the associated shooting cycle among the multiple activity videos. For example, when the fish sample to be identified includes three activity videos (activity video a, activity video b, and activity video c), the server can respectively determine whether the effective duration of activity video a is greater than the shooting cycle, whether the effective duration of activity video b is greater than the shooting cycle, and whether the effective duration of activity video c is greater than the shooting cycle. If the effective durations of activity videos a and b are both greater than the associated shooting cycle, and the effective duration of activity video c is not greater than the associated shooting cycle, the number of IDs is 2.

[0064] Among them, the number of IDs can represent the risk level of the fish samples to be identified. If the number of IDs is larger, the risk level of the fish samples to be identified is higher, and the health hazard to the breeders is greater; if the number of IDs is smaller, the risk level of the fish samples to be identified is lower, and the health hazard to the breeders is smaller.

[0065] S107. If the number of IDs is greater than zero, the server determines the fish sample to be identified as a risky target fish.

[0066] Specifically, when the number of IDs is greater than zero, that is, when there is at least one active video associated with the fish sample to be identified whose effective duration is greater than the shooting cycle, the server will determine the fish sample to be identified as a risky target fish, which can effectively improve breeding safety.

[0067] In one implementation, the fish sample to be identified may also include a target fish ID. After the server collects the number of IDs of the active videos whose effective duration is greater than the associated shooting cycle in multiple active videos, the following steps may be performed: if the number of IDs is equal to zero, the fish sample to be identified is determined as a valid target fish, and the valid target fish is stored in the third plan log set, so that when a feeding request including the target fish ID is obtained, if the target fish ID of the target valid target fish in the third plan log set is consistent with the target fish ID in the feeding request, the feeding request is fed back so that the fish can eat smoothly. In this way, the feed depot system can only allow the fish to swim after obtaining the valid target fish, which can avoid the risk of discovering the target fish after the feed is put into the feed depot, and can effectively improve the safety of breeding.

[0068] In one implementation, after the server determines the fish sample to be identified as a risk target fish, the following steps may be performed: the server determines that the effective duration associated with each activity video in multiple activity videos is greater than the risk activity video of the shooting cycle, and transmits an early warning notification to the mobile device, the early warning notification includes the risk activity video, and the early warning notification is used to warn the mobile device to modify the fish swimming scale and / or the fish order of magnitude of the fish swimming scale associated with the risk activity video in the fish sample to be identified. The mobile device may be a mobile device operated by a salesperson, and by transmitting the early warning notification including the risk activity video to the mobile device used by the salesperson, the salesperson can modify the fish swimming scale and / or the fish order of magnitude of the fish swimming scale associated with the risk activity video based on the risk activity video, which is conducive to improving the safety of breeding.

[0069] In one implementation, the warning notification may be a voice notification, a pop-up notification, or a vibration notification, which is not limited in this embodiment of the present invention.

[0070] In one implementation, in addition to determining whether there is excessive fish swimming in a single target fish (i.e., the fish sample to be identified) (see steps S101 to S107 for details), the server can also determine whether there is a combined excessive fish swimming in multiple target fish.

[0071] In one implementation, each of the multiple target fish may further include a requesting person attribute and a requesting time. After the server collects the number of IDs of the multiple activity videos whose effective duration is greater than the associated shooting cycle, the server may further perform the following steps:

[0072] a. If the number of IDs is zero, the server collects the first target fish in the first preset plan log set whose request person attributes are consistent with the request person attributes of the fish sample to be identified, and the difference between the request time and the request time of the fish sample to be identified is not greater than the preset threshold. That is, the server can collect all the target fish of the same customer recently, and merge all the target fish into a merged target fish, and then determine whether the merged target fish is a risky target fish. In this way, the breeding safety can be improved.

[0073] The request person attribute is used for a unique ID client, and the request person attribute may be an ID number, etc. It should be noted that the preset threshold may be a default setting of the server, or may be determined by the server based on fish operations, which is not limited in the embodiment of the present invention.

[0074] b. The server obtains the valid number of items associated with each of the multiple activity videos of the first target fish based on the mapping value of the fish swimming scale associated with each of the multiple activity videos of the first target fish and the fish order of magnitude associated with the fish swimming scale. It should be noted that the execution process of step b can refer to the specific description in step S102, which is not repeated here.

[0075] c. The server adds the valid number of items associated with each consistent activity video in all consistent activity videos among the multiple activity videos of the first target fish and the multiple activity videos of the fish sample to be identified, and obtains the combined valid number of items associated with each consistent activity video in all consistent activity videos.

[0076] For example, if the valid numbers of items associated with the three activity videos (i.e., activity video a, activity video b, and activity video c) included in the fish sample to be identified are 24, 12, and 3 respectively, and the valid numbers of items associated with the two activity videos (i.e., activity video a and activity video b) included in the first target fish are 12 and 6 respectively, the consistent activity videos are activity video a and activity video b, and the combined valid numbers of items associated with activity video a and activity video b are 36 and 18 respectively.

[0077] d. The server collects the daily shooting duration associated with each consistent activity video in all consistent activity videos in the first planned log set. It should be noted that the execution process of step d can refer to the specific description in step S103, which will not be repeated here.

[0078] e. The ratio between the number of combined valid items associated with each consistent activity video in all consistent activity videos and the daily shooting duration is determined as the combined valid duration associated with each consistent activity video in all consistent activity videos. It should be noted that the execution process of step e can refer to the specific description in step S104, which is not repeated here.

[0079] f. The server collects the shooting period associated with each consistent activity video in all consistent activity videos. It should be noted that the execution process of step f can refer to the specific description in step S105, which will not be repeated here.

[0080] g. The server uses the consistent activity videos whose combined effective duration is greater than the associated shooting period among all consistent activity videos as the first ID, and counts the number of the first IDs.

[0081] The judgment efficiency can be improved by comparing only the combined effective duration and shooting period of the consistent activity videos in the fish sample to be identified and the first target fish, rather than comparing the combined effective duration and shooting period of all activity videos in the fish sample to be identified and the first target fish.

[0082] h. If the number of the first ID is greater than zero, the server determines the fish sample to be identified as a risky target fish.

[0083] Specifically, if the number of the first ID is greater than zero, it indicates that there is a risky purchase behavior for the fish indicated by the attribute of the requesting person. At this time, the server can determine the fish sample to be identified as a risky target fish. In one implementation, the server can also determine the first target fish as a risky target fish, wherein the first target fish can be a judged target fish (i.e., a target fish that has been judged separately) or an unjudged target fish. If the first target fish is an unjudged target fish, then after determining the first target fish as a risky target fish, the server can no longer make a separate judgment on the first target fish, which is beneficial to saving processing resources.

[0084] In one implementation, after the server determines the fish sample to be identified as a risky target fish, it can also determine the requesting person attribute as a risky requesting person attribute, and transmit the risky requesting person attribute to the risk control system, so that the risk control system can add the risky requesting person attribute to the blacklist based on the number of risks associated with the risky requesting person attribute.

[0085] In one implementation, after the server determines the fish sample to be identified as a risk target fish, it can also perform the following steps: determine whether there is a preset activity video in all consistent activity videos, if there is a preset activity video in all consistent activity videos, collect the shooting period associated with the preset activity video, and if the combined effective time associated with the preset activity video is greater than the shooting period, output an alarm notification. In one implementation, the alarm notification can be a voice notification, a pop-up notification, or a vibration notification, which is not limited in the embodiment of the present invention.

[0086] Among them, the preset fish indicated by the preset activity video can be some fish that are too active and will cause great danger to the body of the breeder, such as piranhas. The shooting period associated with the preset activity video means that when the breeder is active for more than the preset fish during the shooting period, it will cause great danger to the breeder's body. For example, when the preset fish is a piranha, the shooting period can be 10 days. When the combined effective time associated with the preset activity video is greater than the shooting period, the output alarm notification can warn the salesperson that the fish indicated by the requesting person's attributes may have an abnormal tendency, so that the fish can be observed in detail, which is conducive to improving breeding safety.

[0087] By implementing the embodiment of the present invention, the effective duration associated with each activity video in the fish sample to be identified can be obtained based on the effective number of associated items and the daily shooting duration of each activity video in the fish sample to be identified, and the effective duration associated with each activity video can be compared with the shooting period. When the comparison result is that the number of IDs of activity videos whose effective duration is greater than the associated shooting period is greater than zero, it indicates that there is at least one activity video indicating an excess of fish in the fish sample to be identified, that is, by implementing the embodiment of the present invention, risky target fish can be identified, which is beneficial to improving breeding safety.

[0088] See also Figure 2 , Figure 2 FIG. 1 is a flow chart of another lightweight fish identification method based on edge intelligence provided by an embodiment of the present invention. Specifically, Figure 2 As shown, another lightweight fish identification method based on edge intelligence in an embodiment of the present invention may include but is not limited to the following steps:

[0089] S201. The server collects fish samples to be identified, where the fish samples to be identified include multiple activity videos, the swimming scale of fish associated with each activity video in the multiple activity videos, and the fish order of magnitude of the swimming scale of the fish.

[0090] It should be noted that the execution process of step S201 can be found in Figure 1 The specific description of step S101 is not repeated here.

[0091] In one implementation, the fish sample to be identified may also include the requester attributes and the request time, wherein the request time refers to the time when the fish sample to be identified is obtained. For example, if Li Si goes to the reservoir to pick fish on May 7, 2022, then the request time is May 7, 2022, and the time when the target fish (the salesperson gives Li Si the same day) is obtained is also May 7, 2022.

[0092] S202: The server determines whether there is plan information associated with the requesting person's attributes in the second plan log set.

[0093] The planning information includes the planning time. The planning time associated with the requester attribute can be the time of the current request determined by the salesperson when the fish indicated by the requester attribute made a historical request. For some breeders who need to feed for a long time, due to the limitation of the nutritional feed for a single fish swim, they need to go to the reservoir regularly to get feed. In order to prevent the breeders from getting feed in advance, the salesperson can specify the time when the breeder will get feed in the future, that is, the planning time, when the breeder makes a historical request.

[0094] In one implementation, the salesperson can determine the planned time based on the nutritional feed prescribed in history. For example, if the salesperson prescribed 30 days of nutritional feed to the breeder when the last request was made, the planned time is the 30th day after the last request time.

[0095] S203: If the second plan log set contains plan information associated with the requesting person's attributes, the server collects the plan time included in the plan information.

[0096] S204. If the request time is earlier than the planned time, the server obtains the valid number of items associated with each of the multiple activity videos based on the fish swimming scale associated with each of the multiple activity videos. The valid number of items associated with each activity video is obtained after tracking and detecting the order of fish of the fish swimming scale associated with the activity video.

[0097] Specifically, if the request time is earlier than the planned time, it indicates that the fish are fed in advance. The reason for feeding the fish in advance may be: a large number of fish are collected at a lower price by feeding in advance. To prevent the above situation, the server can start the judgment program (i.e., trigger the step of obtaining the number of valid items associated with each of the multiple activity videos based on the fish swimming scale associated with each of the multiple activity videos) when it is judged that the request time is earlier than the planned time, and then judge whether there is still an excess of fish swimming in the target fish.

[0098] In one implementation, the fish sample to be identified may also include a first category ID, and the planning information may also include a second category ID. After the server collects the planning time included in the planning information, the following steps may be performed: if the request time is earlier than the planning time, the first category ID and the second category ID are collected; if the first category ID and the second category ID are consistent, a step is triggered based on the fish swimming scale associated with each of the multiple activity videos to obtain the valid number of items associated with each of the multiple activity videos.

[0099] In one implementation, when the reasons for the historical request of the breeder and the current request are different, that is, the request categories are different, it is likely that the request time is earlier than the planned time, but at this time, the request time earlier than the planned time does not belong to early feeding. Therefore, when the request time is earlier than the planned time, by judging whether the first category ID and the second category ID are consistent, the probability that the fish sample to be identified is a risk target fish can be further predicted. If the request time is earlier than the planned time, and the first category ID and the second category ID are inconsistent, it can be considered that the probability that the fish sample to be identified is a risk target fish is low. When the request time is earlier than the planned time, and the first category ID and the second category ID are consistent, it can be considered that the probability that the fish sample to be identified is a risk target fish is high. At this time, the step of obtaining the valid number of items associated with each of the multiple active videos based on the fish swimming scale associated with each active video in the multiple active videos is triggered, that is, the judgment procedure is started, which can improve the safety of breeding.

[0100] S205: The server collects the daily shooting duration associated with each of the multiple activity videos in the first planned log set.

[0101] S206. The server determines a ratio between the number of valid items associated with each of the multiple activity videos and the daily shooting duration as the valid duration associated with each of the multiple activity videos.

[0102] S207: The server collects a shooting period associated with each of the multiple activity videos.

[0103] S208. The server collects the number of IDs of the activity videos whose effective duration is greater than the associated shooting period from the multiple activity videos.

[0104] S209. If the number of IDs is greater than zero, the server determines the fish sample to be identified as a risky target fish.

[0105] It should be noted that the execution process of steps S205 to S209 can be respectively referred to Figure 1 The specific description of steps S103 to S107 is not repeated here.

[0106] By implementing the embodiments of the present invention, when the request time is earlier than the planned time, that is, when it is judged that the fish has pre-feeding behavior, the judgment program can be triggered, and then based on the valid number of associated active videos in the fish sample to be identified and the daily shooting duration, the valid duration associated with each active video in the fish sample to be identified is obtained, and the valid duration associated with each active video is compared with the shooting cycle. When the comparison result is that the number of IDs of active videos whose valid duration is greater than the associated shooting cycle is greater than zero, it indicates that there is at least one active video in the fish sample to be identified that indicates an excess of fish, that is, by implementing the embodiments of the present invention, risky target fish can be judged, which is beneficial to improving breeding safety.

[0107] See also Figure 3 , Figure 3 is a schematic diagram of the structure of a lightweight fish identification system based on edge intelligence provided by an embodiment of the present invention. Specifically, Figure 3 As shown, the lightweight fish identification system based on edge intelligence may include:

[0108] The collection module 301 is used to collect fish samples to be identified, and the fish samples to be identified include multiple activity videos, the swimming scale of fish associated with each activity video in the multiple activity videos, and the fish order of magnitude of the swimming scale of the fish.

[0109] The processing module 302 is used to obtain the valid number of items associated with each of the multiple activity videos based on the fish swimming scale associated with each of the multiple activity videos. The valid number of items associated with each activity video is obtained after tracking and detecting the fish order of magnitude of the fish swimming scale associated with the activity video.

[0110] The collection module 301 is further configured to collect the daily shooting duration associated with each of the multiple activity videos in the first plan log set.

[0111] The processing module 302 is further configured to determine a ratio between the number of valid items associated with each of the multiple activity videos and the daily shooting duration as the valid duration associated with each of the multiple activity videos.

[0112] The collection module 301 is further used to collect the number of IDs of the activity videos whose effective duration is greater than the associated shooting period from among the multiple activity videos.

[0113] The processing module 302 is further configured to determine the fish sample to be identified as a risky target fish if the number of IDs is greater than zero.

[0114] In one implementation, the fish sample to be identified may also include the requesting person's attributes and the requesting time. The processing module 302 may also be used to determine whether there is planning information associated with the requesting person's attributes in the second planning log set; the collection module 301 may also be used to collect the planning time included in the planning information if there is planning information associated with the requesting person's attributes in the second planning log set; the processing module 302 may also be used to trigger the step of obtaining the number of valid items associated with each of the multiple activity videos based on the fish swimming scale associated with each of the multiple activity videos if the request time is earlier than the planned time.

[0115] In one implementation, the fish sample to be identified may also include a first category ID, and the planning information may also include a second category ID. The acquisition module 301 may also be used to collect the first category ID and the second category ID if the request time is earlier than the planned time; the processing module 302 may also be used to trigger a step of obtaining the number of valid items associated with each of the multiple activity videos based on the fish swimming scale associated with each of the multiple activity videos if the first category ID and the second category ID are consistent.

[0116] In one implementation, the acquisition module 301 is specifically used to obtain a judgment operation from a mobile device, the judgment operation includes a fish sample to be identified, and the judgment operation is obtained when the mobile device determines a trigger operation of a judgment control on the fish swimming environment; the processing module 302 can also be used to determine a risk activity video with an effective duration greater than a shooting period associated with each activity video in a plurality of activity videos; in one implementation, the edge intelligence-based lightweight fish identification system 30 can also include a transmission module 303, which is used to transmit an early warning notification to the mobile device, the early warning notification includes a risk activity video, and the early warning notification is used to warn the mobile device to modify the fish swimming scale and / or the order of magnitude of the fish swimming scale associated with the risk activity video in the fish sample to be identified.

[0117] In one implementation, the fish sample to be identified may also include body size, the first planned log set may include multiple log records, and the collection module 301 is specifically used to determine a target log record associated with the body size in the multiple log records, and different log records in the multiple log records store different daily shooting durations associated with each of the multiple activity videos; the daily shooting duration associated with each of the multiple activity videos is collected in the target log record.

[0118] In one implementation, the fish sample to be identified may also include a target fish ID. The processing module 302 may also be used to determine the fish sample to be identified as a valid target fish if the number of IDs is equal to zero; the valid target fish is stored in a third planned log set so that when a feeding request including the target fish ID is obtained, if the target fish ID of the target valid target fish in the third planned log set is consistent with the target fish ID in the feeding request, the feeding request is fed back.

[0119] In one implementation, the processing module 302 is specifically used to collect a mapping value associated with the fish order of magnitude associated with each of the multiple activity videos from the fourth planned log set based on the fish order of magnitude associated with the fish swimming scale of each of the multiple activity videos; multiply the fish swimming scale associated with each of the multiple activity videos by the mapping value associated with the fish order of magnitude of the fish swimming scale to obtain the valid number of items associated with each of the multiple activity videos.

[0120] In the several embodiments provided in the present application, it should be understood that the disclosed systems, systems and methods can be implemented in other ways. For example, the system embodiments described above are only schematic. For example, the division of modules is only a logical function division. There may be other division methods in actual implementation, such as multiple modules or components can be combined or integrated into another system, or some features can be ignored or not executed. Another point is that the mutual coupling or direct coupling or communication connection shown or discussed can be through some interfaces, indirect coupling or communication connection of systems or modules, which can be electrical, mechanical or other forms.

[0121] In addition, each functional module in each embodiment of the present application can be integrated into a processing module, or each module can exist physically separately, or two or more modules can be integrated into one module. The above integrated modules can be implemented in the form of hardware or in the form of software functional modules. The above is only an implementation method of the present application, and does not limit the patent scope of the present application. Any equivalent structure or equivalent process transformation made using the contents of the specification and drawings of this application, or directly or indirectly used in other related technical fields, are also included in the patent protection scope of the present application.

[0122] The specific implementation methods of the invention are described in detail above, but they are only examples, and the present application is not limited to the specific implementation methods described above. For those skilled in the art, any equivalent modification or substitution of the invention is also within the scope of the present application, and therefore, the equalization, modification, and improvement made without departing from the spirit and principle of the present application should be included in the scope of the present application.

Claims

1. A lightweight fish identification method based on edge intelligence, characterized in that: include: Collecting fish samples to be identified, the fish samples to be identified include multiple activity videos, the scale of fish swimming associated with each activity video in the multiple activity videos and the magnitude of the fish swimming scale, the attributes of the requesting person, the requesting time, and the first category ID; Determine whether there is plan information associated with the requesting person attribute in the second plan log set; If the second plan log set contains the plan information associated with the requester attribute, the plan time and the second category ID included in the plan information are collected; the plan time is the time of the future request determined based on the fish indicated by the requester attribute during the historical request; If the request time is earlier than the planned time, and the first category ID is consistent with the second category ID, then based on the fish swimming scale associated with each of the multiple activity videos, the valid number of items associated with each of the multiple activity videos is obtained, where the valid number of items associated with each of the multiple activity videos is obtained by tracking and detecting the fish magnitude of the fish swimming scale associated with the activity video; Collecting the daily shooting duration associated with each of the multiple activity videos in the first plan log set; Determine the ratio between the number of valid items associated with each of the multiple activity videos and the daily shooting duration as the valid duration associated with each of the multiple activity videos; Collecting the shooting period associated with each of the multiple activity videos; the shooting period associated with each activity video refers to the period during which the fish indicated by the activity video can be active at most; Collecting the number of IDs of the multiple activity videos whose effective duration is greater than the associated shooting period; If the number of IDs is greater than zero, the fish sample to be identified is determined as a risky target fish.

2. A lightweight fish identification method based on edge intelligence as claimed in claim 1, characterized in that: The step of collecting fish samples to be identified comprises: Acquire a judgment operation from a mobile device, the judgment operation including the fish sample to be identified, the judgment operation being obtained when the mobile device judges a trigger operation of a judgment control on a fish swimming environment; After determining the fish sample to be identified as a risk target fish, the method further includes: Determine a risk activity video whose effective duration associated with each activity video in the multiple activity videos is greater than a shooting period; An early warning notification is transmitted to the mobile device, the early warning notification including the risk activity video, and the early warning notification is used to warn the mobile device to modify the fish swimming scale associated with the risk activity video in the fish sample to be identified and / or the order of fish in the fish swimming scale.

3. A lightweight fish identification method based on edge intelligence as claimed in claim 1, characterized in that: The fish sample to be identified also includes body size, the first planned log set includes multiple log records, and the daily shooting duration associated with each of the multiple activity videos collected in the first planned log set includes: Determining a target log record associated with the body size from among the plurality of log records, wherein different log records among the plurality of log records store different daily shooting durations associated with each of the plurality of activity videos; The daily shooting duration associated with each of the multiple activity videos is collected in the target log record.

4. The lightweight fish identification method based on edge intelligence according to claim 1, characterized in that: The fish sample to be identified also includes a target fish ID. After collecting the number of IDs of the active videos whose effective duration is greater than the associated shooting cycle in the multiple active videos, the method further includes: If the number of IDs is zero, the fish sample to be identified is determined as a valid target fish; The valid target fish is stored in the third planned log set so that when a feeding request including a target fish ID is obtained, if the target fish ID of a target valid target fish in the third planned log set is consistent with the target fish ID in the feeding request, the feeding request is fed back.

5. The lightweight fish identification method based on edge intelligence according to claim 1, characterized in that: The obtaining, based on the fish swimming scale associated with each of the multiple activity videos, the number of valid items associated with each of the multiple activity videos comprises: Based on the fish magnitude of the fish swimming scale associated with each of the plurality of activity videos, collecting a mapping value associated with the fish magnitude from a fourth planned log set; The fish swimming scale associated with each of the multiple activity videos and the mapping value associated with the fish magnitude of the fish swimming scale are multiplied to obtain the valid number of items associated with each of the multiple activity videos.

6. A lightweight fish identification system based on edge intelligence, characterized in that: The system includes a module for executing a lightweight fish identification method based on edge intelligence as described in any one of claims 1 to 5.

Images