Fish school state monitoring method and related device

Abstract

The invention discloses a fish school state monitoring method and a related device, and relates to the technical field of artificial intelligence, and the method comprises the steps: carrying out the training of a fish school sound signal sample and a fish school state label corresponding to the fish school sound signal sample in advance, and obtaining a fish school state monitoring model; the to-be-monitored fish school sound signal is input into the fish school state monitoring model, a fish school state monitoring result output after the fish school state monitoring model processes the to-be-monitored fish school sound signal can be obtained, and due to the fact that the fish school sound signal can be transmitted underwater in a long distance in the deep and far sea and the equipment requirement is low, the fish school state monitoring result can be obtained. Therefore, the method is suitable for effectively monitoring the fish school state in the deep sea culture environment.

Description

Technical Field

[0001] The present application relates to the field of artificial intelligence technology, and in particular to a fish school status monitoring method and related devices. Background Art

[0002] Fish status monitoring is an important guarantee and necessary condition for fishery farming to avoid risks and obtain high economic benefits. However, deep-sea aquaculture cages are far from the shore, and it is impossible to rely on traditional manual inspection methods to monitor the aquaculture status of fish in the aquaculture cages around the clock.

[0003] At present, the monitoring of fish status mostly adopts computer vision-based methods, which require clear water, sufficient light and stable background as the premise. Therefore, it is often limited to indoor or small farms with good lighting conditions. The lighting conditions cannot be met in deep-sea aquaculture environments, limiting the application effect of such methods.

[0004] Therefore, how to provide a fish status monitoring method suitable for deep-sea aquaculture environments has become a technical problem that needs to be urgently solved by technicians in this field. Summary of the invention

[0005] In view of the above problems, the present application provides a fish status monitoring method and related devices to achieve the purpose of effectively monitoring the status of fish in deep-sea aquaculture environments. The specific scheme is as follows:

[0006] The first aspect of the present application provides a method for monitoring the status of a school of fish, comprising:

[0007] Obtaining sound signals of fish schools to be monitored;

[0008] The fish school sound signal to be monitored is input into a fish school status monitoring model to obtain a fish school status monitoring result output by the fish school status monitoring model after processing the fish school sound signal to be monitored. The fish school status monitoring model is trained using fish school sound signal samples and fish school status labels corresponding to the fish school sound signal samples.

[0009] In a possible implementation, obtaining the sound signal of the school of fish to be monitored includes:

[0010] Obtaining initial fish school sound signals;

[0011] Performing standardization processing on the initial fish school sound signal to obtain a standardized fish school sound signal;

[0012] The fish school sound signal after the standardized processing is the fish school sound signal to be monitored.

[0013] In a possible implementation, the step of performing standardization processing on the initial fish school sound signal to obtain the standardized fish school sound signal includes:

[0014] The initial fish school sound signal is subjected to at least one of data conversion processing, DC removal processing and pre-filtering processing to obtain a standardized fish school sound signal.

[0015] In a possible implementation, performing at least one of data conversion, DC removal and pre-filtering on the initial sound signal to obtain a standardized sound signal includes:

[0016] Performing data conversion processing on the initial fish school sound signal to obtain a fish school sound signal after data conversion processing;

[0017] Performing DC removal processing on the fish school sound signal after the data conversion processing to obtain the fish school sound signal after the DC removal processing;

[0018] The fish school sound signal after the DC removal process is pre-filtered to obtain the fish school sound signal after the standardization process.

[0019] In a possible implementation, the training method of the fish school status monitoring model includes:

[0020] Using the fish school sound signal sample as input to the encoder in the deep autoencoder, with the output of the decoder in the deep autoencoder approaching the fish school sound signal sample as the goal, the deep autoencoder is unsupervisedly trained to obtain a trained deep autoencoder;

[0021] Connecting the encoder in the trained deep autoencoder to a preset classification network to construct the fish school status monitoring model;

[0022] The fish school sound signal sample is input into the encoder in the trained deep autoencoder, and the fish school status monitoring model is trained with the goal that the predicted label of the preset classification network is close to the fish school status label corresponding to the fish school sound signal sample.

[0023] In a possible implementation, the fish school status monitoring model processes the fish school sound signal to be monitored and outputs the fish school status monitoring result, including:

[0024] The fish school state monitoring model extracts and compresses the fish school sound signal to be monitored to obtain a feature vector of the fish school sound signal to be monitored;

[0025] The characteristic vectors of the fish school sound signals to be monitored are classified, and the fish school status monitoring results are output.

[0026] A second aspect of the present application provides a fish school status monitoring device, comprising:

[0027] An acquisition unit, used for acquiring the sound signal of the school of fish to be monitored;

[0028] The monitoring unit is used to input the fish school sound signal to be monitored into a fish school status monitoring model to obtain a fish school status monitoring result output by the fish school status monitoring model after processing the fish school sound signal to be monitored. The fish school status monitoring model is trained using fish school sound signal samples and fish school status labels corresponding to the fish school sound signal samples.

[0029] In a possible implementation, the acquiring unit includes:

[0030] An initial signal acquisition unit, used for acquiring an initial fish school sound signal;

[0031] A standardization processing unit, used for performing standardization processing on the initial fish school sound signal to obtain a standardized fish school sound signal;

[0032] The fish school sound signal after the standardized processing is the fish school sound signal to be monitored.

[0033] In a possible implementation, the standardization processing unit is specifically used to:

[0034] The initial fish school sound signal is subjected to at least one of data conversion processing, DC removal processing and pre-filtering processing to obtain a standardized fish school sound signal.

[0035] In a possible implementation, the standardization processing unit is specifically used to:

[0036] Performing data conversion processing on the initial fish school sound signal to obtain a fish school sound signal after data conversion processing;

[0037] Performing DC removal processing on the fish school sound signal after the data conversion processing to obtain the fish school sound signal after the DC removal processing;

[0038] The fish school sound signal after the DC removal process is pre-filtered to obtain the fish school sound signal after the standardization process.

[0039] In a possible implementation, the device further includes: a fish school status monitoring model training unit, wherein the fish school status monitoring model training unit is specifically used to:

[0040] Using the fish school sound signal sample as input to the encoder in the deep autoencoder, with the output of the decoder in the deep autoencoder approaching the fish school sound signal sample as the goal, the deep autoencoder is unsupervisedly trained to obtain a trained deep autoencoder;

[0041] Connecting the encoder in the trained deep autoencoder to a preset classification network to construct the fish school status monitoring model;

[0042] The fish school sound signal sample is input into the encoder in the trained deep autoencoder, and the fish school status monitoring model is trained with the goal that the predicted label of the preset classification network is close to the fish school status label corresponding to the fish school sound signal sample.

[0043] In a possible implementation, the monitoring unit is specifically configured to:

[0044] The fish school state monitoring model extracts and compresses the fish school sound signal to be monitored to obtain a feature vector of the fish school sound signal to be monitored;

[0045] The characteristic vectors of the fish school sound signals to be monitored are classified, and the fish school status monitoring results are output.

[0046] The third aspect of the present application provides a computer program product, including computer-readable instructions. When the computer-readable instructions are executed on an electronic device, the electronic device implements the fish school status monitoring method of the first aspect or any implementation of the first aspect.

[0047] A fourth aspect of the present application provides an electronic device, comprising at least one processor and a memory connected to the processor, wherein:

[0048] The memory is used to store computer programs;

[0049] The processor is used to execute the computer program so that the electronic device can implement the fish school status monitoring method of the above-mentioned first aspect or any implementation manner of the first aspect.

[0050] A fifth aspect of the present application provides a computer-readable storage medium, which carries one or more computer programs. When the one or more computer programs are executed by an electronic device, the electronic device can implement the fish school status monitoring method of the above-mentioned first aspect or any implementation method of the first aspect.

[0051] By means of the above technical scheme, the present application provides a fish school status monitoring method and related devices, which pre-train a fish school status monitoring model using fish school sound signal samples and fish school status labels corresponding to the fish school sound signal samples. After obtaining the fish school sound signal to be monitored, the fish school sound signal to be monitored is input into the fish school status monitoring model, and the fish school status monitoring result output by the fish school status monitoring model after processing the fish school sound signal to be monitored can be obtained. Since the fish school sound signal can be transmitted long distances underwater in the deep sea and the equipment requirements are relatively low, it can be suitable for effective monitoring of the fish school status in the deep sea aquaculture environment. BRIEF DESCRIPTION OF THE DRAWINGS

[0052] The above and other features, advantages and aspects of the embodiments of the present disclosure will become more apparent with reference to the following detailed description in conjunction with the accompanying drawings. Throughout the accompanying drawings, the same or similar reference numerals represent the same or similar elements. It should be understood that the drawings are schematic and the originals and elements are not necessarily drawn to scale.

[0053] Figure 1 A schematic diagram of the structure of a fish school status monitoring system provided in an embodiment of the present application;

[0054] Figure 2 A schematic diagram of a flow chart of a fish school status monitoring method provided in an embodiment of the present application;

[0055] Figure 3 A flow chart of a fish school status monitoring model training method provided in an embodiment of the present application;

[0056] Figure 4 A schematic diagram of a deep autoencoder training provided in an embodiment of the present application;

[0057] Figure 5 A schematic diagram of the structure of a fish school status monitoring device provided in an embodiment of the present application;

[0058] Figure 6 A schematic diagram of the structure of an electronic device provided in an embodiment of the present application. DETAILED DESCRIPTION

[0059] The following describes the embodiments of the present application in conjunction with the drawings in the embodiments of the present application. The terms used in the implementation method section of the present application are only used to explain the specific embodiments of the present application, and are not intended to limit the present application.

[0060] The embodiments of the present application are described below in conjunction with the accompanying drawings. Those skilled in the art will appreciate that, with the development of technology and the emergence of new scenarios, the technical solutions provided in the embodiments of the present application are also applicable to similar technical problems.

[0061] The terms "first", "second", etc. in the specification and claims of the present application and the above-mentioned drawings are used to distinguish similar objects, and need not be used to describe a specific order or sequential order. It should be understood that the terms used in this way can be interchangeable under appropriate circumstances, which is only to describe the distinction mode adopted by the objects of the same attributes when describing in the embodiments of the present application. In addition, the terms "including" and "having" and any of their variations are intended to cover non-exclusive inclusions, so that the process, method, system, product or equipment comprising a series of units need not be limited to those units, but may include other units that are not clearly listed or inherent to these processes, methods, products or equipment.

[0062] In response to the national "blue" economic strategic needs for the construction of marine ranches, my country's fishery farming is transforming and upgrading towards the development direction of modern, offshore, and large-scale smart fisheries. With the continuous development of technology and the gradual upgrading of equipment, my country's fishery farming technology has made certain breakthroughs on the basis of the traditional "watching the stall and guarding the business", but the current development of smart fisheries is only at the stage of digital monitoring, and there is still a certain gap from the realization of intelligent and smart deep-sea aquaculture models.

[0063] Fish status monitoring is an important guarantee and necessary condition for fishery farming to avoid risks and obtain high economic benefits. However, deep-sea aquaculture cages are far from the shore, and it is impossible to rely on traditional manual inspection methods to monitor the aquaculture status of fish in the aquaculture cages around the clock.

[0064] At present, the monitoring of fish status mostly adopts computer vision-based methods, and the implementation steps are generally as follows: first, fish school images are obtained through underwater cameras to form a fish school image dataset; second, the images in the fish school image dataset are preprocessed; then, features are extracted from the preprocessed images; then, the fish school status features are extracted using optical flow methods and the like; then, the normalized features of each image are used as feature vectors for training to obtain a fish school status monitoring model; finally, the fish school status monitoring model is evaluated to obtain the final fish school status monitoring model.

[0065] The fish status monitoring method based on computer vision is based on computer vision image recognition technology, which relies on underwater cameras to capture, process, model and identify the behavior of fish schools in the breeding box. Its main disadvantages are: on the one hand, underwater cameras have high requirements for lighting conditions, and require relatively clean water and a clean imaging background, etc., which cannot be met in deep-sea breeding boxes, resulting in extremely limited observation distance and range of underwater cameras; on the other hand, it has high requirements for equipment. First, the underwater camera must have self-cleaning capabilities, otherwise it cannot maintain clear image capture capabilities for a long time. Secondly, due to the large amount of high-definition image data, the data storage and transmission burden is heavy, both of which greatly increase the equipment cost burden.

[0066] It can be seen that the fish status monitoring method based on computer vision requires clear water, sufficient light and stable background as the premise. Therefore, it is often limited to indoor or small farms with good lighting conditions. The lighting conditions cannot be met for deep-sea aquaculture environments, which limits the application effect of such methods.

[0067] In order to provide a fish status monitoring method suitable for deep-sea aquaculture environments, technical personnel in this field have found that many studies have discovered and confirmed that farmed fish can produce sound through swim bladder vibration, bone friction and underwater movement, and sound waves are the only signal form that can be transmitted over long distances underwater. Compared with underwater visual signals, the propagation of sound signals is not restricted by lighting conditions, and the propagation distance and range are farther. Therefore, it can be suitable for monitoring the status of fish in deep-sea aquaculture boxes. In addition, compared with underwater cameras and other equipment, passive acoustic monitoring equipment has the advantages of low deployment requirements, lower power consumption, and low data acquisition and storage requirements. This provides a new way for the development of fish status monitoring technology for deep-sea aquaculture platforms.

[0068] Based on this, the embodiment of the present application provides a fish status monitoring system, which monitors the status of fish in deep-sea aquaculture boxes by collecting, processing and analyzing the fish behavior and environmental sound signals in the deep-sea aquaculture boxes.

[0069] Reference Figure 1 , Figure 1 This is a schematic diagram of the structure of a fish school status monitoring system provided in an embodiment of the present application. Figure 1 As shown, the fish status monitoring system includes a deep-sea aquaculture platform located in the deep sea and a data monitoring and management center located on shore.

[0070] In a possible implementation, the deep-sea aquaculture platform includes aquaculture boxes and underwater acoustic equipment, and the underwater acoustic equipment includes N underwater front-end devices (such as hydrophones), signal measurement amplifiers, signal collectors, and data transmission and storage devices. It should be noted that when configuring the underwater acoustic equipment, the layout plan can be determined based on the specific geometric dimensions of the aquaculture box and the parameters of the underwater acoustic equipment to ensure that the range of the underwater front-end equipment (such as hydrophones) can cover the entire range of the aquaculture box, that is, it needs to meet the following requirements:

[0071]

[0072] in, represents the range of the nth underwater front-end device (e.g., hydrophone), For the entire breeding box range.

[0073] Specifically, the underwater acoustic equipment collects sound signals in the breeding box and transmits the collected sound signals to the data monitoring and management center via wireless data transmission. The data monitoring and management center processes and analyzes the received sound signals to monitor the status of fish schools in deep-sea breeding boxes.

[0074] In order to process and analyze the received sound signals, the inventors of the present application first constructed and trained a fish school status monitoring model by combining artificial intelligence deep learning technology, and then deployed the fish school status monitoring model to the data monitoring and management center. The data monitoring and management center can process and analyze the received sound signals through the fish school status monitoring model.

[0075] Based on the above-mentioned fish school status monitoring system, an embodiment of the present application provides a fish school status monitoring method, which can be executed by a data monitoring and management center.

[0076] The following is a detailed introduction to the fish school status monitoring method of the embodiment of the present application in conjunction with the accompanying drawings.

[0077] Reference Figure 2 , Figure 2 A flow chart of a fish school status monitoring method provided in an embodiment of the present application is shown as follows: Figure 2 As shown, a fish school status monitoring method provided in an embodiment of the present application may include steps S101 to S102, and these steps are described in detail below.

[0078] S101: Acquire the sound signal of the fish school to be monitored;

[0079] In a possible implementation, obtaining the sound signal of the school of fish to be monitored includes:

[0080] Acquire an initial fish school sound signal; and use the initial fish school sound signal as the fish school sound signal to be monitored.

[0081] In the present application, the initial fish school sound signal is a sound signal collected by underwater acoustic equipment in a deep-sea aquaculture platform and transmitted to a data monitoring and management center.

[0082] The initial fish sound signal is generated by the fish to be monitored. Its source may include typical daily activities such as feeding, swimming, and resting, as well as atypical activities such as overactive fish or insufficient fish activity. It is used to indicate the aquaculture status of fish in deep-sea aquaculture platforms.

[0083] Considering that the initial fish sound signals collected by underwater acoustic equipment in deep-sea aquaculture platforms may be inconsistent in form or have interference, it will affect the accuracy of fish status monitoring results.

[0084] In order to solve the above problems, the present application provides another implementation method for obtaining the sound signal of the fish school to be monitored, which is as follows:

[0085] Obtaining initial fish school sound signals;

[0086] Performing standardization processing on the initial fish school sound signal to obtain a standardized fish school sound signal;

[0087] The fish school sound signal after the standardized processing is the fish school sound signal to be monitored.

[0088] As an implementable embodiment, the standardization processing of the initial fish school sound signal to obtain the standardized fish school sound signal includes: performing at least one of data conversion processing, DC removal processing and pre-filtering processing on the initial fish school sound signal to obtain the standardized fish school sound signal.

[0089] Among them, in this application, data conversion processing refers to data format conversion processing, the purpose of which is to unify the fish school sound signals of different formats (such as bin format, dat format, wav format, etc.) output by different underwater front-end devices into the same format (such as wav format). In this application, the purpose of DC removal processing is to remove the DC offset in the fish school sound signal. In this application, the purpose of pre-filtering processing is to remove out-of-band noise and interference in the fish school sound signal.

[0090] As an optimal implementation method, in the present application, the initial fish school sound signal can be sequentially subjected to data conversion processing, DC removal processing and pre-filtering processing to obtain a standardized fish school sound signal.

[0091] Specifically, the initial fish school sound signal is first subjected to data conversion processing to obtain a fish school sound signal after data conversion processing; then, the fish school sound signal after data conversion processing is subjected to DC removal processing to obtain a fish school sound signal after DC removal processing; finally, the fish school sound signal after DC removal processing is subjected to pre-filtering processing to obtain the fish school sound signal after standardization processing.

[0092] For ease of understanding, in a possible implementation, the following formula may be used for DC removal:

[0093]

[0094] Among them, x represents the sound signal of the school of fish after data conversion and processing, there are M points in total, x m is the mth point, mean(x) is the mean amplitude of the time domain signal, x rc This is the fish school sound signal after DC removal.

[0095] In a possible implementation, the following formula may be used for pre-filtering:

[0096] in, represents the pre-filter, The configured filter parameters need to be determined according to the noise frequency band of the specific farmed fish species.

[0097] S102: Input the fish school sound signal to be monitored into a fish school status monitoring model to obtain a fish school status monitoring result output by the fish school status monitoring model after processing the fish school sound signal to be monitored, wherein the fish school status monitoring model is trained using fish school sound signal samples and fish school status labels corresponding to the fish school sound signal samples.

[0098] In this application, according to the purpose of monitoring the acoustic environment in the breeding box, the state of the fish school can be divided into normal state and abnormal state in general. The sources of the sound signals generated by the fish school in the normal state include typical daily activities such as feeding, swimming, and resting. The sources of the sound signals generated by the fish school in the abnormal state include atypical activities such as the fish school being too active, the fish school being insufficiently active, etc. The reasons may be drastic changes in the breeding box environment or being attacked, changes in the health status of the fish school, etc. In this application, the abnormal state is the focus of fish school status monitoring. After the fish school state is detected to be abnormal, the fish school state can be intervened in time.

[0099] Then, in the present application, the fish school status monitoring results output by the fish school status monitoring model include a normal state and an abnormal state.

[0100] The fish school sound signal samples may include sound signals generated by the fish school in a normal state and sound signals generated by the fish school in an abnormal state. The sound signals generated by the fish school in a normal state and the sound signals generated by the fish school in an abnormal state may be initial signals or signals after standardized processing, preferably signals after standardized processing. The fish school state label corresponding to the sound signal generated by the fish school in a normal state may be a normal state, and the fish school state label corresponding to the sound signal generated by the fish school in an abnormal state may be an abnormal state.

[0101] In a possible implementation, in an actual breeding environment, sound signals generated by normal activities of fish schools with a total duration of T1 hours and sound signals generated by abnormal activities of fish schools with a total duration of T2 hours can be collected as fish school sound signal samples. According to the probability distribution of events, the following conditions must be met:

[0102]

[0103] It should be noted that in the present application, there is no limitation on the structure and training method of the fish school status monitoring model. As long as the model is trained using fish school sound signal samples and the fish school status labels corresponding to the fish school sound signal samples, the model used to monitor the fish school status is within the protection scope of the present application.

[0104] The present embodiment provides a method for monitoring the status of a school of fish. A school of fish status monitoring model is trained in advance using fish sound signal samples and fish status labels corresponding to the fish sound signal samples. After obtaining the sound signal of the school of fish to be monitored, the sound signal of the school of fish to be monitored is input into the fish status monitoring model, so that the school of fish status monitoring result output by the fish status monitoring model after processing the sound signal of the school of fish to be monitored can be obtained. Since the sound signal of the school of fish can be transmitted underwater for a long distance in the deep sea and the equipment requirements are relatively low, the school of fish status can be effectively monitored in the deep sea aquaculture environment.

[0105] In order to facilitate the understanding of the fish school status monitoring model, a specific structure and training method of the fish school status monitoring model are described in detail below.

[0106] In a possible implementation, in this application, a fish state monitoring model can be constructed and trained based on a deep auto encoder (DAE) and a convolutional neural network (CNN). A deep auto encoder is a special type of neural network composed of multiple layers of neural networks. The core idea is to compress the input data into a low-dimensional latent space through an encoder, and then reconstruct the representation in the latent space back to the original data space through a decoder.

[0107] Based on this, the present application discloses a training method for a fish school status monitoring model, referring to Figure 3 , Figure 3 A flow chart of a fish school status monitoring model training method provided in an embodiment of the present application is as follows: Figure 3 As shown, the method may include the following steps:

[0108] S301: using the fish school sound signal sample to input the encoder in the deep autoencoder, with the output of the decoder in the deep autoencoder approaching the fish school sound signal sample as the goal, performing unsupervised training on the deep autoencoder to obtain a trained deep autoencoder;

[0109] For ease of understanding, assume that the fish school sound signal sample is ,in, , L is the number of samples, M is the number of points of each training sample sound signal data, and the encoding matrix is ​​obtained after passing through the encoder in the deep autoencoder , where h is the number of neurons in the last layer of the encoder:

[0110]

[0111] In order to achieve feature dimensionality reduction, h<M is selected.

[0112] Next, the encoding matrix is ​​decoded by the deep autoencoder Decode and reconstruct the X input:

[0113]

[0114] After reconstruction and The dimensions need to be consistent. Figure 4 shown.

[0115] In the unsupervised training process of the deep autoencoder, it is necessary to minimize the input and The square error determines the coding part With the decoding part Parameters:

[0116]

[0117] S302: Connecting the encoder in the trained deep autoencoder to a preset classification network to construct the fish school status monitoring model;

[0118] In the present application, the encoder and decoder in the trained deep autoencoder can be disassembled, and the disassembled encoder is connected to a preset classification network to construct the fish school status monitoring model.

[0119] In a possible implementation, ResNet-18 composed of CNN can be used as a preset classification network.

[0120] S303: Input the fish school sound signal sample into the encoder in the trained deep autoencoder, and train the fish school status monitoring model with the goal of making the output of the preset classification network close to the fish school status label corresponding to the fish school sound signal sample.

[0121] In the present application, the fish school sound signal sample is input into the encoder of the trained deep autoencoder, and the prediction label of the preset classification network is close to the fish school state label corresponding to the fish school sound signal sample. When the fish school state monitoring model is trained, the loss can adopt the cross entropy loss:

[0122]

[0123] in, Represents the fish school status label corresponding to the fish school sound signal sample, Represents the predicted label of the preset classification network.

[0124] According to the calculated loss, the network parameters of the fish school status monitoring model are adjusted through feedback until the model converges to the allowable error range, and finally a trained fish school status monitoring model is obtained.

[0125] As an implementable method, the fish school status monitoring model processes the fish school sound signal to be monitored and outputs the fish school status monitoring result, which may include:

[0126] The fish school state monitoring model extracts and compresses the fish school sound signal to be monitored to obtain a feature vector of the fish school sound signal to be monitored;

[0127] The characteristic vectors of the fish school sound signals to be monitored are classified, and the fish school status monitoring results are output.

[0128] For ease of understanding, assume that the eigenvector of the fish school sound signal to be monitored is , and use the L×h dimensional data as the input of ResNet-18. After the data dimension is adjusted, the 7×7 convolution layer is first connected to the ResNet network, and the convolution layer is connected to the batch normalization layer and then the 3×3 maximum pooling layer. Subsequently, 4 modules composed of residual blocks are used, and each module uses 2 residual blocks with the same number of output channels. The number of channels of the first module is the same as the number of input channels. In each subsequent module, the number of channels of the previous module is doubled in the first residual block, and the height and width are halved. Finally, the number of output channels is changed to 2 through the fully connected layer, and the predicted probabilities of normal and abnormal states are output. The fish school status monitoring results can be obtained based on the predicted probabilities of normal and abnormal states.

[0129] A fish school status monitoring method provided in an embodiment of the present application is introduced above, and a device for executing the above-mentioned fish school status monitoring method will be introduced below.

[0130] See also Figure 5 , Figure 5This is a schematic diagram of the structure of a fish school status monitoring device provided in an embodiment of the present application. Figure 5 As shown, the fish school status monitoring device comprises:

[0131] An acquisition unit 11 is used to acquire the sound signal of the school of fish to be monitored;

[0132] The monitoring unit 12 is used to input the fish school sound signal to be monitored into the fish school status monitoring model to obtain the fish school status monitoring result output by the fish school status monitoring model after processing the fish school sound signal to be monitored. The fish school status monitoring model is trained using fish school sound signal samples and fish school status labels corresponding to the fish school sound signal samples.

[0133] In a possible implementation, the acquiring unit includes:

[0134] An initial signal acquisition unit, used for acquiring an initial fish school sound signal;

[0135] A standardization processing unit, used for performing standardization processing on the initial fish school sound signal to obtain a standardized fish school sound signal;

[0136] The fish school sound signal after the standardized processing is the fish school sound signal to be monitored.

[0137] In a possible implementation, the standardization processing unit is specifically used to:

[0138] The initial fish school sound signal is subjected to at least one of data conversion processing, DC removal processing and pre-filtering processing to obtain a standardized fish school sound signal.

[0139] In a possible implementation, the standardization processing unit is specifically used to:

[0140] Performing data conversion processing on the initial fish school sound signal to obtain a fish school sound signal after data conversion processing;

[0141] Performing DC removal processing on the fish school sound signal after the data conversion processing to obtain the fish school sound signal after the DC removal processing;

[0142] The fish school sound signal after the DC removal process is pre-filtered to obtain the fish school sound signal after the standardization process.

[0143] In a possible implementation, the device further includes: a fish school status monitoring model training unit, wherein the fish school status monitoring model training unit is specifically used to:

[0144] Using the fish school sound signal sample as input to the encoder in the deep autoencoder, with the output of the decoder in the deep autoencoder approaching the fish school sound signal sample as the goal, the deep autoencoder is unsupervisedly trained to obtain a trained deep autoencoder;

[0145] Connecting the encoder in the trained deep autoencoder to a preset classification network to construct the fish school status monitoring model;

[0146] The fish school sound signal sample is input into the encoder in the trained deep autoencoder, and the fish school status monitoring model is trained with the goal that the predicted label of the preset classification network is close to the fish school status label corresponding to the fish school sound signal sample.

[0147] In a possible implementation, the monitoring unit is specifically configured to:

[0148] The fish school state monitoring model extracts and compresses the fish school sound signal to be monitored to obtain a feature vector of the fish school sound signal to be monitored;

[0149] The characteristic vectors of the fish school sound signals to be monitored are classified, and the fish school status monitoring results are output.

[0150] The present application also provides an electronic device in an embodiment. Figure 6 As shown, it shows a schematic diagram of the structure of an electronic device suitable for implementing the embodiment of the present application. The electronic device in the embodiment of the present application may include but is not limited to fixed terminals such as mobile phones, laptops, PDAs (personal digital assistants), PADs (tablet computers), desktop computers, etc. Figure 6 The electronic device shown is merely an example and should not bring any limitation to the functions and scope of use of the embodiments of the present application.

[0151] like Figure 6 As shown, the electronic device may include a processing device (e.g., a central processing unit, a graphics processing unit, etc.) 601, which can perform various appropriate actions and processes according to a program stored in a read-only memory (ROM) 602 or a program loaded from a storage device 608 to a random access memory (RAM) 603. When the electronic device is powered on, various programs and data required for the operation of the electronic device are also stored in the RAM 603. The processing device 601, the ROM 602, and the RAM 603 are connected to each other via a bus 604. An input / output (I / O) interface 605 is also connected to the bus 604.

[0152] Typically, the following devices may be connected to the I / O interface 605: an input device 606 including, for example, a touch screen, a touch pad, a keyboard, a mouse, a camera, a microphone, an accelerometer, a gyroscope, etc.; an output device 607 including, for example, a liquid crystal display (LCD), a speaker, a vibrator, etc.; a storage device 608 including, for example, a memory card, a hard disk, etc.; and a communication device 609. The communication device 609 may allow the electronic device to communicate with other devices wirelessly or by wire to exchange data. Although Figure 6 An electronic device having various devices is shown, but it should be understood that it is not required to implement or possess all the devices shown. More or fewer devices may be implemented or possessed instead.

[0153] An embodiment of the present application also provides a computer program product, including computer-readable instructions. When the computer-readable instructions are executed on an electronic device, the electronic device implements any of the fish school status monitoring methods provided in the embodiments of the present application.

[0154] A computer-readable storage medium is also provided in an embodiment of the present application. The storage medium carries one or more computer programs. When the one or more computer programs are executed by an electronic device, the electronic device can implement any of the fish school status monitoring methods provided in the embodiments of the present application.

[0155] It should also be noted that the device embodiments described above are merely schematic, wherein the units described as separate components may or may not be physically separated, and the components displayed as units may or may not be physical units, that is, they may be located in one place, or they may be distributed over multiple network units. Some or all of the modules may be selected according to actual needs to achieve the purpose of the scheme of this embodiment. In addition, in the drawings of the device embodiments provided by the present application, the connection relationship between the modules indicates that there is a communication connection between them, which may be specifically implemented as one or more communication buses or signal lines.

[0156] Through the description of the above implementation mode, the technicians in the field can clearly understand that the present application can be implemented by means of software plus necessary general hardware, and of course, it can also be implemented by special hardware including special integrated circuits, special CPUs, special memories, special components, etc. In general, all functions completed by computer programs can be easily implemented by corresponding hardware, and the specific hardware structure used to implement the same function can also be various, such as analog circuits, digital circuits or special circuits. However, for the present application, software program implementation is a better implementation mode in more cases. Based on such an understanding, the technical solution of the present application is essentially or the part that contributes to the prior art can be embodied in the form of a software product, which is stored in a readable storage medium, such as a computer floppy disk, a U disk, a mobile hard disk, a ROM, a RAM, a disk or an optical disk, etc., including a number of instructions to enable a computer device (which can be a personal computer, a training device, or a network device, etc.) to execute the methods described in each embodiment of the present application.

[0157] In the above embodiments, all or part of the embodiments may be implemented by software, hardware, firmware or any combination thereof. When implemented by software, all or part of the embodiments may be implemented in the form of a computer program product.

[0158] The computer program product includes one or more computer instructions. When the computer program instructions are loaded and executed on a computer, the process or function described in the embodiment of the present application is generated in whole or in part. The computer may be a general-purpose computer, a special-purpose computer, a computer network, or other programmable devices. The computer instructions may be stored in a computer-readable storage medium, or transmitted from one computer-readable storage medium to another computer-readable storage medium. For example, the computer instructions may be transmitted from a website site, a computer, a training device, or a data center by wired (e.g., coaxial cable, optical fiber, digital subscriber line (DSL)) or wireless (e.g., infrared, wireless, microwave, etc.) mode to another website site, computer, training device, or data center. The computer-readable storage medium may be any available medium that a computer can store or a data storage device such as a training device, a data center, etc. that includes one or more available media integrations. The available medium may be a magnetic medium, (e.g., a floppy disk, a hard disk, a tape), an optical medium (e.g., a DVD), or a semiconductor medium (e.g., a solid-state drive (SSD)), etc.

Claims

1. A method for monitoring the status of a school of fish, characterized in that: include: Obtaining sound signals of fish schools to be monitored; The fish school sound signal to be monitored is input into a fish school status monitoring model to obtain a fish school status monitoring result output by the fish school status monitoring model after processing the fish school sound signal to be monitored. The fish school status monitoring model is trained using fish school sound signal samples and fish school status labels corresponding to the fish school sound signal samples.

2. The method according to claim 1, characterized in that: The step of obtaining the fish school sound signal to be monitored comprises: Obtaining initial fish school sound signals; Performing standardization processing on the initial fish school sound signal to obtain a standardized fish school sound signal; The fish school sound signal after the standardized processing is the fish school sound signal to be monitored.

3. The method according to claim 2, characterized in that The step of performing standardization processing on the initial fish school sound signal to obtain the standardized fish school sound signal comprises: The initial fish school sound signal is subjected to at least one of data conversion processing, DC removal processing and pre-filtering processing to obtain a standardized fish school sound signal.

4. The method according to claim 3, characterized in that The step of performing at least one of data conversion, DC removal and pre-filtering on the initial sound signal to obtain a standardized sound signal includes: Performing data conversion processing on the initial fish school sound signal to obtain a fish school sound signal after data conversion processing; Performing DC removal processing on the fish school sound signal after the data conversion processing to obtain the fish school sound signal after the DC removal processing; The fish school sound signal after the DC removal process is pre-filtered to obtain the fish school sound signal after the standardization process.

5. The method according to claim 1, characterized in that The training method of the fish school status monitoring model includes: Using the fish school sound signal sample as input to the encoder in the deep autoencoder, with the output of the decoder in the deep autoencoder approaching the fish school sound signal sample as the goal, the deep autoencoder is unsupervisedly trained to obtain a trained deep autoencoder; Connecting the encoder in the trained deep autoencoder to a preset classification network to construct the fish school status monitoring model; The fish school sound signal sample is input into the encoder in the trained deep autoencoder, and the fish school status monitoring model is trained with the goal that the predicted label of the preset classification network is close to the fish school status label corresponding to the fish school sound signal sample.

6. The method according to claim 1, characterized in that The fish school status monitoring model processes the fish school sound signal to be monitored and outputs the fish school status monitoring result, including: The fish school state monitoring model extracts and compresses the fish school sound signal to be monitored to obtain a feature vector of the fish school sound signal to be monitored; The characteristic vectors of the fish school sound signals to be monitored are classified, and the fish school status monitoring results are output.

7. A fish school status monitoring device, characterized in that: include: An acquisition unit, used for acquiring the sound signal of the school of fish to be monitored; The monitoring unit is used to input the fish school sound signal to be monitored into a fish school status monitoring model to obtain a fish school status monitoring result output by the fish school status monitoring model after processing the fish school sound signal to be monitored. The fish school status monitoring model is trained using fish school sound signal samples and fish school status labels corresponding to the fish school sound signal samples.

8. A computer program product, characterized in that It comprises computer-readable instructions, and when the computer-readable instructions are executed on an electronic device, the electronic device implements the fish school status monitoring method as claimed in any one of claims 1 to 6.

9. An electronic device, characterized in that: The method comprises at least one processor and a memory connected to the processor, wherein: The memory is used to store computer programs; The processor is used to execute the computer program so that the electronic device can implement the fish school status monitoring method as described in any one of claims 1 to 6.

10. A computer-readable storage medium, characterized in that: The storage medium carries one or more computer programs, and when the one or more computer programs are executed by an electronic device, the electronic device can implement the fish school status monitoring method as described in any one of claims 1 to 6.

Images