Intelligent fish protector fish weighing and returning system based on image recognition

Abstract

The invention discloses an intelligent fish protector fish weighing and returning system based on image recognition, which relates to the technical field of image recognition and comprises a control terminal, a fish protector module, an image recognition module, a weighing module and a fish returning module. The fish basket module comprises a first fish basket and a second fish basket, the first fish basket is used for storing water for the fished fish temporarily, the second fish basket is used for storing the fish with the weight reaching the standard, and the image recognition module captures images of the fish stored in the first fish basket, recognizes each fish according to the images, generates recognition tag information of each fish and sends the recognition tag information to the second fish basket; the weighing module weighs each fish according to the identification label information of each fish so as to obtain weight information of each fish, the weight information is combined with the identification label information to generate a final identity identification voucher of each fish, and the control terminal receives the identity identification voucher to construct a control instruction and sends the control instruction to the fish returning module; and the fish returning module decides to release or throw the fish in the first fish basket according to the control instruction.

Description

Technical Field

[0001] The invention relates to the technical field of image recognition, in particular to an intelligent fish guard, fish weighing and fish returning system based on image recognition. Background Art

[0002] In traditional fishing activities, the handling of fish caught is usually cumbersome and lacks effective management methods. Generally, anglers need to temporarily store the fish they catch in a fish basket, then weigh and sort them according to their size, and then select the fish that meet the weight requirements. However, this process often requires manual operation, which is not only inefficient but also prone to errors. In addition, due to the lack of effective identification and recording methods, it is difficult for anglers to accurately record the identity information of each fish, which to some extent affects the management of fish. Summary of the invention

[0003] In order to solve the above problems, the object of the present invention is to provide an intelligent fish guard, fish weighing and fish returning system based on image recognition.

[0004] The object of the present invention can be achieved by the following technical solutions: An intelligent fish guard, fish weighing and fish returning system based on image recognition, comprising a control terminal, wherein the control terminal is communicatively connected with a fish guard module, an image recognition module, a weighing module and a fish returning module;

[0005] The fish guard module includes a fish guard 1 and a fish guard 2, the fish guard 1 and the fish guard 2 are connected to each other through a pipe network channel, and the pipe network channel is provided with an opening and closing device, and the opening and closing device is used to select opening or closing of the pipe network channel;

[0006] The fish guard is used to store water temporarily for the fish caught;

[0007] The second fish guard is used to store fish that meet the weight standard. When the storage volume of the fish in the second fish guard reaches the upper volume limit, the fish in the second fish guard is taken out;

[0008] The image recognition module is used to capture the image of the fish stored in the fish guard, identify each fish according to the image, generate identification tag information of each fish, and transmit the identification tag information to the weighing module;

[0009] The weighing module is used to weigh each fish according to the identification tag information of each fish, thereby obtaining the weight information of each fish, combining the weight information of each fish with the identification tag information, and generating the final identity certificate of each fish;

[0010] The control terminal receives all the identity identification credentials, constructs control instructions according to the identity identification credentials, and sends all the control instructions to the fish return module;

[0011] The fish return module is used to decide whether to release the fish entering the pipe network channel in fish guard one according to the control instruction. If so, the weight of the fish released in fish guard one meets the standard and enters fish guard two. If not, it is decided whether to release the fish in fish guard one.

[0012] Furthermore, the process of the image recognition module capturing the image of the fish stored in the fish guard includes:

[0013] A high-resolution camera is arranged in the above-water part of the fish guard. Each fish in the fish guard is tracked, captured and photographed by the high-resolution camera, and then several images of each fish are taken. The imaging quality of each image of the same fish is judged in turn whether it meets the requirements. The image with the highest imaging quality among the images that meet the requirements of each fish is selected as the final image to be identified, so as to obtain the final image of each fish.

[0014] Furthermore, the process of identifying each fish according to the image, generating identification tag information of each fish, and transmitting the identification tag information to the weighing module includes:

[0015] Perform grayscale processing on the image of each fish, convert each fish image into a corresponding grayscale image, construct a grayscale co-occurrence matrix based on the grayscale image, and extract the corresponding image texture information after analyzing the grayscale co-occurrence matrix;

[0016] Obtaining an appearance image corresponding to each known type of fish, processing the appearance image, and extracting appearance feature information corresponding to each known type of fish, the appearance feature information including overall body shape features, fin features, head features, and body structure features of the fish;

[0017] Construct a matching pool, input the image texture information extracted from the image of each fish into the matching pool, integrate the appearance feature information of all known types of fish as a data set to be matched, and input the data set to be matched into the matching pool, and then match the image texture information with the appearance feature information in the matching pool;

[0018] If the image texture information records the overall body shape, fin features, head features, and body structure features of a known type of fish, the image texture information successfully matches the appearance feature information of the known type of fish, and the type of fish corresponding to the current known type is marked as label element one, and the size of the fish is identified according to the image texture information as label element two;

[0019] Otherwise, the match fails and no action is performed;

[0020] Integrate tag element one and tag element two to serve as identification tag information of each fish;

[0021] An information transmission channel is constructed between the image recognition module and the weighing module, and all identification tag information is transmitted through the information transmission channel.

[0022] Furthermore, the weighing module weighs each fish according to the identification tag information of each fish, thereby obtaining the weight information of each fish, and combining the weight information of each fish with the identification tag information to generate the final identity identification certificate of each fish. The process includes:

[0023] The weighing module is composed of a data reading device, a guiding device, a weighing device and a data management terminal;

[0024] The data reading device is used to read the identification tag information of each fish, and then establish an identity account for each fish according to the identification tag information;

[0025] The guiding device is physically connected to the weighing device, the guiding device is provided with a sensing area, and the guiding device guides the fish in the sensing area to the weighing device;

[0026] When the weighing device detects that a fish is loaded on it, the weighing device records the weight information of the current fish in real time and enters it into the identity account corresponding to the current fish. The identity account combines the weight information and identification tag information of each fish to generate the final identity certificate of each fish;

[0027] The data management terminal is used to store the identity account of each fish, and to perform data security protection on all identity accounts in the data management terminal.

[0028] Furthermore, the data security protection process includes:

[0029] A data processing node is constructed for each identity account association. The data processing node is used to review and determine the access request information of the current identity account. If the review determines that the corresponding access request information in the current data processing node contains illegal text fragments, the current access is prohibited;

[0030] If the audit determines that the access request information corresponding to the current data processing node does not contain any illegal text fragments, the current access is allowed;

[0031] Record all operations on the identity account associated with each data processing node monitored, and then build an operation record log for the identity account corresponding to each fish. The operation record log records the time, operation content and operation object of each operation on the identity account.

[0032] Furthermore, the process of the control terminal receiving all the identity identification credentials and constructing a control instruction according to the identity identification credentials includes:

[0033] A communication period is set, during which the control terminal receives all identity accounts in the data management terminal, and then obtains the identity identification credentials recorded in each identity account;

[0034] Constructing a control instruction according to the identity identification credential, the control instruction including instruction one and instruction two;

[0035] If the corresponding weight information recorded in the identity identification credential shows that the weight of the fish meets the standard, then instruction one is constructed to be associated with the current identity identification credential. If the corresponding weight information recorded in the identity identification credential shows that the weight of the fish does not meet the standard, then instruction two is constructed to be associated with the current identity identification credential, and all control instructions constructed in the control terminal are transmitted to the fish return module.

[0036] Furthermore, the process in which the fish return module decides whether to release the fish entering the pipe network channel in the fish guard according to the control instruction includes:

[0037] When the fish return module receives instruction 1, it controls the weighing device in the fish guard 1 to rise to the same level as the pipe network channel, transfers the fish on the weighing device that meet the weight standard into the pipe network channel, configures the working parameters of the opening and closing device, transmits the working parameters to the fish guard module, and the fish guard module controls the opening and closing device according to the working parameters to open the pipe network channel, thereby releasing the fish in the pipe network channel in the fish guard 1 into the fish guard 2;

[0038] When the fish return module receives instruction two, it controls the weighing device to perform a flipping operation, returns the fish loaded on the weighing device to the fish guard one, and counts the number of flipping operations, which is recorded as S, where S is an integer greater than 0.

[0039] Furthermore, the content of deciding whether to release the fish of Yuhuyi is as follows:

[0040] An upper threshold value of the number of fish stored in the fish guard 1 is set, and the upper threshold value is recorded as S′, where S′ is an integer greater than 0, and the numerical value relationship between S and S′ is compared to determine whether to release the fish stored in the fish guard 1;

[0041] When S≥S′, the fish in the fish guard 1 are released into the water. When S<S′, no operation is performed on the fish guard 1.

[0042] Compared with the prior art, the beneficial effects of the present invention are as follows: in the fish guard module, fish guard one is used for temporarily storing caught fish, and fish guard two is used for storing fish that meet the weight standard. This graded storage method helps to improve the efficiency of fish screening. The introduction of the image recognition module can capture the image of the fish in fish guard one, and generate identification label information of each fish according to the image, thereby realizing automatic identification of fish and reducing the error rate of manual identification. The weighing module weighs the fish according to the identification label information, and the generated identity identification certificate contains the weight information of the fish, providing a unique identity for each fish, which is convenient for subsequent management and traceability; the control terminal receives the identity identification certificate and constructs a control instruction, and the fish return module decides whether to release the fish to fish guard two or to release the fish that does not meet the weight standard into the water according to the control instruction, thereby realizing the management of fish, reducing the complexity of manual operation, and helping to protect fish resources and improve the scientific nature of fishery management. BRIEF DESCRIPTION OF THE DRAWINGS

[0043] Figure 1 It is a schematic diagram of the present invention. DETAILED DESCRIPTION

[0044] like Figure 1 As shown, an intelligent fish guard, fish weighing and fish returning system based on image recognition includes a control terminal, wherein the control terminal is communicatively connected with a fish guard module, an image recognition module, a weighing module and a fish returning module;

[0045] The fish guard module includes a fish guard 1 and a fish guard 2, the fish guard 1 and the fish guard 2 are connected to each other through a pipe network channel, and the pipe network channel is provided with an opening and closing device, and the opening and closing device is used to select opening or closing of the pipe network channel;

[0046] The fish guard is used to store water temporarily for the fish caught;

[0047] The second fish guard is used to store fish that meet the weight standard. When the storage volume of the fish in the second fish guard reaches the upper volume limit, the fish in the second fish guard is taken out;

[0048] The image recognition module is used to capture the image of the fish stored in the fish guard, identify each fish according to the image, generate identification tag information of each fish, and transmit the identification tag information to the weighing module;

[0049] The weighing module is used to weigh each fish according to the identification tag information of each fish, thereby obtaining the weight information of each fish, combining the weight information of each fish with the identification tag information, and generating the final identity certificate of each fish;

[0050] The control terminal receives all the identity identification credentials, constructs control instructions according to the identity identification credentials, and sends all the control instructions to the fish return module;

[0051] The fish return module is used to decide whether to release the fish entering the pipe network channel in fish guard one according to the control instruction. If so, the weight of the fish released in fish guard one meets the standard and enters fish guard two. If not, it is decided whether to release the fish in fish guard one.

[0052] It should be further explained that, in a specific implementation process, the process of the image recognition module capturing the image of the fish stored in the fish guard includes:

[0053] A high-resolution camera is arranged in the above-water part of the fish guard. The high-resolution camera has IP68 waterproof function. Each fish in the fish guard is tracked, captured and photographed by the high-resolution camera, and then several images of each fish are taken. The imaging quality of each image of the same fish is judged in turn whether it meets the requirements. The image with the highest imaging quality among the images that meet the requirements of each fish is selected as the final image to be identified, so as to obtain the final image of each fish.

[0054] The contents for judging whether the image quality meets the requirements are as follows:

[0055] Obtain imaging parameters of fish images captured by a high-resolution camera;

[0056] The imaging parameters include image resolution, image contrast and signal-to-noise ratio, and the values ​​of image resolution, image contrast and signal-to-noise ratio are respectively recorded as d1, d2 and d3;

[0057] When the image meets the imaging quality requirement, the lower thresholds of image resolution, image contrast and signal-to-noise ratio are set, and the lower thresholds are recorded as Fb, Db and Xz respectively;

[0058] When d1≥Fb, d2≥Db and d3≥Xz, it is judged that the imaging quality of the image meets the requirements;

[0059] Otherwise, it is judged that the imaging quality of the image does not meet the requirements.

[0060] It should be further explained that, in a specific implementation process, the process of identifying each fish according to the image, generating identification tag information of each fish, and transmitting the identification tag information to the weighing module includes:

[0061] Perform grayscale processing on the image of each fish, and then convert each fish image into a corresponding grayscale image, construct a corresponding grayscale co-occurrence matrix based on the grayscale image of each fish, and analyze the grayscale co-occurrence matrix to extract the corresponding image texture information;

[0062] The image texture information is used to characterize the appearance features corresponding to the current fish;

[0063] Obtaining the appearance image corresponding to each known type of fish, and extracting the appearance feature information corresponding to each known type of fish through grayscale processing, constructing a grayscale symbiosis matrix, and analyzing the grayscale symbiosis matrix, the appearance feature information including the overall body shape features, fin features, head features, and body structure features of the fish;

[0064] Construct a matching pool, input the image texture information extracted from the image of each fish into the matching pool, integrate the appearance feature information of all known types of fish as a data set to be matched, and input the data set to be matched into the matching pool, and then match the image texture information with the appearance feature information in the matching pool;

[0065] If the image texture information records the overall body shape, fin features, head features, and body structure features of a known type of fish, the image texture information successfully matches the appearance feature information of the known type of fish, and the type of fish corresponding to the current known type is marked as label element one, and the size of the fish is identified according to the image texture information as label element two;

[0066] Otherwise, the match fails and no operation is performed;

[0067] Integrate tag element one and tag element two to serve as identification tag information of each fish;

[0068] An information transmission channel is constructed between the image recognition module and the weighing module, and the transmission rate of the information transmission channel is monitored in real time. All identification tag information is transmitted through the information transmission channel. When the transmission rate does not meet the preset numerical range, an early warning prompt is generated, and then the bandwidth of the information transmission channel is changed to adjust the transmission rate of the information transmission channel.

[0069] It should be further explained that, in the specific implementation process, the weighing module weighs each fish according to the identification tag information of each fish, and then obtains the weight information of each fish, combines the weight information of each fish with the identification tag information, and generates the final identity identification certificate of each fish. The process includes:

[0070] The weighing module is composed of a data reading device, a guiding device, a weighing device and a data management terminal;

[0071] The data reading device is used to read the identification tag information of each fish, and then establish an identity account for each fish according to the identification tag information, and the information content initially stored in the identity account is an empty string;

[0072] The guiding device is physically connected to the weighing device, and the guiding device is provided with a sensing area. When the fish in the fish guard enters the sensing area of ​​the guiding device, the guiding device guides the fish in the sensing area to the weighing device;

[0073] When the weighing device detects that a fish is loaded on it, the weighing device records the weight information of the current fish in real time and enters the weight information into the identity account corresponding to the current fish. The identity account combines the weight information and identification tag information of each fish to generate the final corresponding identity identification certificate for each fish;

[0074] The data management terminal is used to store the identity account of each fish, and to perform data security protection on all identity accounts in the data management terminal;

[0075] The contents of data security protection are as follows:

[0076] A data processing node is constructed for each identity account association. The data processing node is used to review and determine the access request information of the current identity account. If the review determines that the corresponding access request information in the current data processing node contains illegal text fragments, the current access is prohibited;

[0077] Enter the IP address of the current access object into the preset access blacklist, and perform vulnerability scanning on the data processing node. Through vulnerability scanning, locate several system vulnerabilities in the current data processing node and repair all system vulnerabilities;

[0078] If the audit determines that the access request information corresponding to the current data processing node does not contain any illegal text fragments, the current access is allowed;

[0079] Record all operations on the identity account associated with each data processing node monitored, and then build an operation record log for the identity account corresponding to each fish. The operation record log records the time, operation content and operation object of each operation on the identity account.

[0080] It should be noted that illegal text fragments include text fragments corresponding to virus information, malicious code injection, information attack intrusion, and operation texts for performing data tampering. The text fragments corresponding to the above operations are illegal.

[0081] It should be further explained that, in a specific implementation process, the control terminal receives all identity identification credentials, and the process of constructing a control instruction according to the identity identification credentials includes:

[0082] A communication period is set and recorded as Tc, Tc = [t1, t2], where t1 is the start time of the communication period and t2 is the end time of the communication period. During the communication period, the control terminal receives all identity accounts in the data management terminal, and then obtains the identity identification credentials recorded in each identity account;

[0083] Constructing a control instruction according to the identity identification credential, the control instruction including instruction one and instruction two;

[0084] If the corresponding weight information recorded in the identity identification credential indicates that the weight of the fish meets the standard, the control instruction constructed is instruction one, and instruction one is associated with the current identity identification credential; if the corresponding weight information recorded in the identity identification credential indicates that the weight of the fish does not meet the standard, the control instruction constructed is instruction two, and instruction two is associated with the current identity identification credential;

[0085] Different control instructions correspond to different instruction identifiers;

[0086] The instruction identifier corresponding to instruction one is denoted as Point1;

[0087] The instruction identifier corresponding to instruction 2 is recorded as Point2;

[0088] All control instructions constructed in the control terminal are transmitted to the fish return module.

[0089] It should be further explained that, in the specific implementation process, the process in which the fish return module decides whether to release the fish entering the pipe network channel in the fish guard according to the control instruction includes:

[0090] When the fish return module receives instruction 1, it controls the weighing device in the fish guard 1 to rise to the same level as the pipe network channel, transfers the fish on the weighing device that meet the weight standard into the pipe network channel, configures the working parameters corresponding to the opening and closing device, transmits the working parameters to the fish guard module, and the fish guard module controls the opening and closing device according to the working parameters to open the pipe network channel, thereby releasing the fish in the pipe network channel in the fish guard 1 into the fish guard 2;

[0091] The working parameters include the opening and closing duration, opening and closing rate, and opening and closing size of the pipe network channel;

[0092] Among them, the pipe network channel between Yuhu 1 and Yuhu 2 was initially closed;

[0093] When the fish return module receives instruction two, it controls the weighing device to perform a flipping operation, returns the fish loaded on the weighing device to the fish guard one, and counts the number of flipping operations, which is recorded as S, where S is an integer greater than 0.

[0094] It should be noted that the weighing device is in the shape of a tray. When the fish enters the weighing device, the weighing device detects whether the weight of the currently loaded fish meets the standard. If so, the weighing device is raised and the subsequent release of the fish is executed. If not, the weighing device is flipped. The flipping operation is to flip the weighing device 180 degrees so that the fish loaded thereon leaves the weighing device.

[0095] It should be further explained that in the specific implementation process, the content of deciding whether to release the fish of Yuhuyi is as follows:

[0096] An upper threshold value of the number of fish stored in the fish guard 1 is set, and the upper threshold value is recorded as S′, where S′ is an integer greater than 0, and the numerical value relationship between S and S′ is compared to determine whether to release the fish stored in the fish guard 1;

[0097] When S≥S′, the fish in the fish guard 1 are released into the water. When S<S′, no operation is performed on the fish guard 1.

[0098] The above embodiments are only used to illustrate the technical method of the present invention rather than to limit it. Although the present invention has been described in detail with reference to the preferred embodiments, those skilled in the art should understand that the technical method of the present invention may be modified or replaced by equivalents without departing from the spirit and scope of the technical method of the present invention.

Claims

1. An intelligent fish guard and fish return system based on image recognition, comprising a control terminal, characterized in that: The control terminal is communicatively connected with a fish protection module, an image recognition module, a weighing module and a fish return module; The fish guard module includes a fish guard 1 and a fish guard 2, the fish guard 1 and the fish guard 2 are connected to each other through a pipe network channel, and the pipe network channel is provided with an opening and closing device, and the opening and closing device is used to select opening or closing of the pipe network channel; The fish guard is used to store water temporarily for the fish caught; The second fish guard is used to store fish that meet the weight standard. When the storage volume of the fish in the second fish guard reaches the upper volume limit, the fish in the second fish guard is taken out; The image recognition module is used to capture the image of the fish stored in the fish guard, identify each fish according to the image, generate identification tag information of each fish, and transmit the identification tag information to the weighing module; The weighing module is used to weigh each fish according to the identification tag information of each fish, thereby obtaining the weight information of each fish, combining the weight information of each fish with the identification tag information, and generating the final identity certificate of each fish; The control terminal receives all the identity identification credentials, constructs control instructions according to the identity identification credentials, and sends all the control instructions to the fish return module; The fish return module is used to decide whether to release the fish entering the pipe network channel in fish guard one according to the control instruction. If so, the weight of the fish released in fish guard one meets the standard and enters fish guard two. If not, it is decided whether to release the fish in fish guard one.

2. The intelligent fish guard and fish return system based on image recognition according to claim 1 is characterized in that: The process of the image recognition module capturing the image of the fish stored in the fish guard comprises: A high-resolution camera is arranged in the above-water part of the fish guard. Each fish in the fish guard is tracked, captured and photographed by the high-resolution camera, and then several images of each fish are taken. The imaging quality of each image of the same fish is judged in turn whether it meets the requirements. The image with the highest imaging quality among the images that meet the requirements of each fish is selected as the final image to be identified, so as to obtain the final image of each fish.

3. The intelligent fish guard and fish return system based on image recognition according to claim 2 is characterized in that: The process of identifying each fish based on the image, generating identification tag information for each fish, and transmitting the identification tag information to the weighing module includes: Perform grayscale processing on the image of each fish, convert each fish image into a corresponding grayscale image, construct a grayscale co-occurrence matrix based on the grayscale image, and extract the corresponding image texture information after analyzing the grayscale co-occurrence matrix; Obtaining an appearance image corresponding to each known type of fish, processing the appearance image, and extracting appearance feature information corresponding to each known type of fish, the appearance feature information including overall body shape features, fin features, head features, and body structure features of the fish; Construct a matching pool, input the image texture information extracted from the image of each fish into the matching pool, integrate the appearance feature information of all known types of fish as a data set to be matched, and input the data set to be matched into the matching pool, and then match the image texture information with the appearance feature information in the matching pool; If the image texture information records the overall body shape, fin features, head features, and body structure features of a known type of fish, the image texture information successfully matches the appearance feature information of the known type of fish, and the type of fish corresponding to the current known type is marked as label element one, and the size of the fish is identified according to the image texture information as label element two; Otherwise, the match fails and no action is performed; Integrate tag element one and tag element two to serve as identification tag information of each fish; An information transmission channel is constructed between the image recognition module and the weighing module, and all identification tag information is transmitted through the information transmission channel.

4. The intelligent fish guard and fish return system based on image recognition according to claim 3 is characterized in that: The weighing module weighs each fish according to the identification tag information of each fish, thereby obtaining the weight information of each fish, combining the weight information of each fish with the identification tag information, and generating the final identity identification certificate of each fish. The process includes: The weighing module is composed of a data reading device, a guiding device, a weighing device and a data management terminal; The data reading device is used to read the identification tag information of each fish, and then establish an identity account for each fish according to the identification tag information; The guiding device is physically connected to the weighing device, the guiding device is provided with a sensing area, and the guiding device guides the fish in the sensing area to the weighing device; When the weighing device detects that a fish is loaded on it, the weighing device records the weight information of the current fish in real time and enters it into the identity account corresponding to the current fish. The identity account combines the weight information and identification tag information of each fish to generate the final identity certificate of each fish; The data management terminal is used to store the identity account of each fish, and to perform data security protection on all identity accounts in the data management terminal.

5. The intelligent fish guard and fish return system based on image recognition according to claim 4 is characterized in that: The data security protection process includes: A data processing node is constructed for each identity account association. The data processing node is used to review and determine the access request information of the current identity account. If the review determines that the corresponding access request information in the current data processing node contains illegal text fragments, the current access is prohibited; If the audit determines that the access request information corresponding to the current data processing node does not contain any illegal text fragments, the current access is allowed; Record all operations on the identity account associated with each data processing node monitored, and then build an operation record log for the identity account corresponding to each fish. The operation record log records the time, operation content and operation object of each operation on the identity account.

6. The intelligent fish guard and fish return system based on image recognition according to claim 5 is characterized in that: The process of the control terminal receiving all the identity identification credentials and constructing a control instruction according to the identity identification credentials includes: A communication period is set, during which the control terminal receives all identity accounts in the data management terminal, and then obtains the identity identification credentials recorded in each identity account; Constructing a control instruction according to the identity identification credential, the control instruction including instruction one and instruction two; If the corresponding weight information recorded in the identity identification credential shows that the weight of the fish meets the standard, then instruction one is constructed to be associated with the current identity identification credential. If the corresponding weight information recorded in the identity identification credential shows that the weight of the fish does not meet the standard, then instruction two is constructed to be associated with the current identity identification credential, and all control instructions constructed in the control terminal are transmitted to the fish return module.

7. The intelligent fish guard and fish return system based on image recognition according to claim 6 is characterized in that: The process in which the fish return module decides whether to release the fish entering the pipe network channel in the fish guard according to the control instruction includes: When the fish return module receives instruction 1, it controls the weighing device in the fish guard 1 to rise to the same level as the pipe network channel, transfers the fish on the weighing device that meet the weight standard into the pipe network channel, configures the working parameters of the opening and closing device, transmits the working parameters to the fish guard module, and the fish guard module controls the opening and closing device according to the working parameters to open the pipe network channel, thereby releasing the fish in the pipe network channel in the fish guard 1 into the fish guard 2; When the fish return module receives instruction two, it controls the weighing device to perform a flipping operation, returns the fish loaded on the weighing device to the fish guard one, and counts the number of flipping operations, which is recorded as S, where S is an integer greater than 0.

8. The intelligent fish guard and fish return system based on image recognition according to claim 7 is characterized in that: The following factors determine whether to release fish from Yuhuyi: An upper threshold value of the number of fish stored in the fish guard 1 is set, and the upper threshold value is recorded as S′, where S′ is an integer greater than 0, and the numerical value relationship between S and S′ is compared to determine whether to release the fish stored in the fish guard 1; When S≥S′, the fish in the fish guard 1 are released into the water. When S<S′, no operation is performed on the fish guard 1.

Images