A green peacock ecological monitoring method and system based on singing

Abstract

The present invention discloses a method and system for monitoring the ecology of green peacocks based on their chirping. The method comprises: first constructing a target dataset; the target dataset includes a green peacock chirping dataset, a bird chirping dataset, an interference sound dataset, and an environmental sound dataset; then constructing a deep learning network model based on the target dataset; then collecting sound data from a target monitoring area to obtain target sound data; then extracting sound information from the target sound data to obtain target information; and then identifying the target information using the deep learning network model to obtain a sound recognition result; the sound recognition result includes the chirping of green peacocks, various bird chirping sounds, interference sounds, and environmental sounds; and finally, collecting green peacock ecological data based on the sound recognition results. The embodiments of the present invention can improve the accuracy of green peacock ecological monitoring and can be widely applied in the field of acoustic signal processing technology.

Description

Technical Field

[0001] The present invention relates to the technical field of sound signal processing, and in particular to a green peacock ecology monitoring method and system based on sound. Background Art

[0002] The Green Peafowl, currently found only in Yunnan Province, is a critically endangered species and has been designated a nationally protected wildlife species. They primarily inhabit tropical and subtropical evergreen broad-leaved and mixed forests below 2,000 meters above sea level, and particularly prefer open areas such as sparse woodlands, grasslands, and riverbanks.

[0003] To promote the conservation of the green peafowl and its habitat, it is crucial to monitor its distribution, behavior, and dynamics. However, the green peafowl's alert nature makes systematic manual monitoring difficult.

[0004] Deploying infrared cameras to monitor green peacocks is a relatively advanced technology that minimizes disturbance to the birds. However, the cameras' monitoring range is limited and affected by their viewing angle. Calling is an important form of communication for green peacocks, and their vocalization increases significantly during the breeding season. Furthermore, the green peacock's call is loud and can travel long distances, making it highly feasible to monitor the birds through its call.

[0005] Passive bird monitoring technology based on bird calls involves deploying recorders in the monitoring area. Recordings are recorded on a set schedule and stored on SD cards, which are retrieved by staff at regular intervals. Machine learning (deep learning) is used to build bird call recognition models to identify the bird species corresponding to the recorded data. Some products also support real-time transmission of recorded data via 4G wireless. However, there are no specific monitoring methods or system solutions for the green peacock's call.

[0006] Therefore, the related prior art has the following disadvantages:

[0007] 1. In existing bird monitoring methods based on song, due to the lack of labeled data, deep learning recognition models generally do not support the green peacock species, or the recognition rate of the green peacock species is low.

[0008] 2. Existing bird monitoring methods based on bird calls only output the identified bird species information and do not perform ecological information statistics. Summary of the Invention

[0009] In view of this, an embodiment of the present invention provides a green peacock ecology monitoring method and system based on sound, which can improve monitoring accuracy and efficiency.

[0010] One aspect of an embodiment of the present invention provides a method for monitoring the ecology of green peacocks based on their singing, comprising the following steps:

[0011] Constructing a target data set; the target data set includes a green peacock sound data set, a bird sound data set, an interference sound data set, and an environmental sound data set;

[0012] Constructing a deep learning network model based on the target data set; the deep learning network model includes a sound event detection model, a bird sound recognition model, and a green peacock sound recognition model;

[0013] Collect sound data from the target monitoring area to obtain target sound data;

[0014] Extracting sound information from the target sound data to obtain target information, and identifying the target information using the deep learning network model to obtain a sound recognition result; the sound recognition result includes the sound of a green peacock, various bird calls, interference sounds, and environmental sounds;

[0015] Green peacock ecological data statistics are performed based on the sound recognition results.

[0016] In some embodiments, constructing the target dataset comprises the following steps:

[0017] Deploy a monophonic recording device in areas where green peacocks are likely to appear. The microphone of the monophonic recording device is omnidirectional, has an operating frequency band of 200 Hz-16,000 Hz, a sampling rate of 32,000 Hz or higher, and a sampling accuracy of 16 bits.

[0018] Based on the audio data collected by the monophonic recording device, the green peacock's call type is identified and the behavioral information corresponding to the call is annotated to generate label data for each type of green peacock's call;

[0019] Based on the public information on the public platform, determine the list of bird species in the monitored area and obtain the song data of these bird species;

[0020] Screening out various types of interference sounds based on the audio data collected by the monophonic recording device, including: car driving sounds, car horns, dog barking, human voices, and construction sounds, and annotating the screened interference sound data to generate interference sound label data;

[0021] Download rain, wind, and thunder sound data from open source websites to construct an environmental sound dataset;

[0022] The target data set is obtained by summarizing the label data of each type of green peacock sound, the sound data of bird species, the label data of interference sounds, and the environmental sound data set.

[0023] In some embodiments, constructing a deep learning network model based on the target data set includes the following steps:

[0024] All types of green peacock calls are considered as an acoustic event. Each type of environmental sound is considered as an acoustic event. Each type of interference sound is considered as an acoustic event. Some bird sound data is selected and combined together to form a bird sound event. The pre-trained logmelBase-Frame-Level model is used as the initial sound event detection model. The collected acoustic event detection dataset is used to retrain the logmelBase-Frame-Level model to obtain the final sound event detection model.

[0025] Each type of green peacock song dataset and various types of bird song datasets were sliced into 2s slices and randomly divided into training set, validation set and test set in a ratio of 7:2:1. The bird sound recognition model and the green peacock song recognition model were trained.

[0026] In some embodiments, retraining the logmelBase-Frame-Level model with the collected acoustic event detection dataset to obtain a final sound event detection model comprises the following steps:

[0027] In data preprocessing, the input signal is first subjected to Fourier transform to obtain the spectrum of the input signal. Then, the power spectrum density is calculated and the power spectrum is processed through a Mel filter. The transfer function of the Mel filter is defined as:

[0028]

[0029] Where H(mk) represents the frequency response function of the Mel filter; F center[m] and F width[m] are the center frequency and width of the mth filter respectively;

[0030] Take the logarithm of the Mel filter processing result to get V[m], V[m] represents the output of the m-th Mel filter group, and serialize the processed Mel filter output V[m] into a feature vector, which is used as the input of the logmelBase-Frame-Level model;

[0031] Based on the features extracted by the Log-Mel process, an embedded representation at the segment level or frame level is obtained;

[0032] The embedding representation of the query set is multiplied by the initialized weight matrix, and the prediction result is obtained through the softmax function, thereby obtaining a sound event detection model. The prediction output includes four types of sound event detection results: green peacock sound, environmental sound, interference sound, and bird sound.

[0033] Among them, the network structure of the sound event detection model includes 2 BasicBlock layers, 2 CNN layers and 1 linear layer.

[0034] In some embodiments, the data sets of each type of green peacock song and various types of bird song data sets are sliced into 2s slices and randomly divided into training set, validation set and test set in a ratio of 7:2:1, and the bird sound recognition model and the green peacock song recognition model are trained, including the following steps:

[0035] The data set is filtered using a bandpass filtering method to obtain a signal y[n] after filtering; wherein, the filtering adopts a window function method to implement a bandpass filter, and the window function adopts a Hanning window;

[0036] Perform Chirplet transformation on the filtered signal to obtain a Chirplet speech graph;

[0037] Among them, Chirplet transformation first requires defining Chirplet atoms. A Chirplet atom can be expressed as: Where f0 represents the center frequency; α represents the frequency modulation rate; t0 represents the center time of the atom; represents a window function, which is 1 from time t0-Δ to t0+Δ, otherwise it is 0; ψ[n] represents a Chirplet atomic operator;

[0038] After the Chirplet atom library is constructed, for each atom in the y[n] sequence obtained through filtering and the atom library Calculate discrete Chirplet transform coefficients The calculation expression is: in, is the complex conjugation of the atom;

[0039] The Chirplet graph is constructed by taking the modulus of the calculated Chirplet transformation coefficients;

[0040] The obtained Chirplet spectrogram is input into the VGGish model for feature extraction. After feature extraction in the VGGish model, the feature data is input into the XGBoost algorithm for classification, and the corresponding probability of each type of green peacock call and the probability of each bird species are output.

[0041] In some embodiments, performing green peacock ecological data statistics based on the sound recognition results includes the following steps:

[0042] The green peacock's calls were counted over time by monitoring area, and the frequency of calls at different times of the day, as well as seasonal variations, were analyzed to generate rhythmic diagrams showing the green peacock's call activity patterns in each monitoring area.

[0043] The various types of green peacock calls are counted by time, and the collected sound data and corresponding labels will be subjected to time series analysis. The frequency of occurrence of each sound type will be counted and associated with time to determine the frequency of green peacock calls in specific time periods. The time when human interference sounds occur will also be counted to analyze the impact of interference sounds on green peacock activities.

[0044] In some embodiments, the performing of green peacock ecological data statistics based on the sound recognition results further includes the following steps:

[0045] Statistical analysis was used to evaluate the relationship between the intensity and frequency of interfering sounds and the singing patterns of green peacocks, and to analyze the effects of interfering sounds on the singing behavior and daily activities of green peacocks.

[0046] Calculate the sound pressure level and spectral components of human interference sounds, and analyze the impact of the interference sound size and frequency components on the behavior of green peacocks;

[0047] The potential effects of sound intensity on the behavior of green peacocks were assessed by analyzing the sound pressure levels of audio clips measured by a sound level meter;

[0048] The frequency distribution of the sound was measured by a spectrum analyzer to analyze the effects of sounds in different frequency ranges on the behavior of the green peacock;

[0049] Observe and count other bird species coexisting in the monitoring area, and analyze the interactions between coexisting bird species.

[0050] Another aspect of the present invention provides a green peacock ecology monitoring system based on singing, comprising:

[0051] The first module is used to construct a target data set; the target data set includes a green peacock sound data set, a bird sound data set, an interference sound data set, and an environmental sound data set;

[0052] The second module is used to build a deep learning network model based on the target data set; the deep learning network model includes a sound event detection model, a bird sound recognition model, and a green peacock sound recognition model;

[0053] The third module is used to collect sound data from the target monitoring area to obtain target sound data;

[0054] A fourth module is configured to extract sound information from the target sound data to obtain target information, and identify the target information using the deep learning network model to obtain a sound recognition result; the sound recognition result includes the sound of a green peacock, various bird calls, interference sounds, and environmental sounds;

[0055] The fifth module is used to collect ecological data of green peacocks based on the sound recognition results.

[0056] Another aspect of an embodiment of the present invention further provides an electronic device, including a processor and a memory;

[0057] The memory is used to store programs;

[0058] The processor executes the program to implement the method described above.

[0059] Another aspect of the present invention provides a computer-readable storage medium, wherein the storage medium stores a program, and the program is executed by a processor to implement the method described above.

[0060] The present invention also discloses a computer program product or computer program, which includes computer instructions stored in a computer-readable storage medium. A processor of a computer device can read the computer instructions from the computer-readable storage medium and execute the computer instructions, causing the computer device to perform the above method.

[0061] The embodiment of the present invention first constructs a target dataset; the target dataset includes a green peacock singing dataset, a bird sound dataset, an interference sound dataset, and an environmental sound dataset; then, a deep learning network model is constructed based on the target dataset; the deep learning network model includes a sound event detection model, a bird sound recognition model, and a green peacock singing recognition model; then, sound data is collected from the target monitoring area to obtain target sound data; acoustic information is extracted from the target sound data to obtain target information, and the target information is recognized using the deep learning network model to obtain a sound recognition result; the sound recognition result includes the green peacock singing, various bird sounds, interference sounds, and environmental sounds; finally, green peacock ecological data statistics are generated based on the sound recognition results. The embodiment of the present invention can improve the accuracy of green peacock ecological monitoring. BRIEF DESCRIPTION OF THE DRAWINGS

[0062] In order to more clearly illustrate the technical solutions in the embodiments of the present invention, the following briefly introduces the drawings required for use in the description of the embodiments. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying any creative work.

[0063] Figure 1 A schematic diagram of an implementation environment provided for an embodiment of the present invention;

[0064] Figure 2 An overall step flow chart provided for an embodiment of the present invention;

[0065] Figure 3 A schematic diagram of the logmelBase-Frame-Level model framework provided in an embodiment of the present invention;

[0066] Figure 4 A flowchart of the process of forming a Chirplet graph provided by an embodiment of the present invention;

[0067] Figure 5 This is a flowchart of the bird song network training provided by an embodiment of the present invention. DETAILED DESCRIPTION

[0068] In order to make the objects, technical solutions and advantages of the present invention more clearly understood, the present invention is further described in detail below with reference to the accompanying drawings and examples. It should be understood that the specific embodiments described herein are only used to explain the present invention and are not intended to limit the present invention. When the following description refers to the drawings, unless otherwise indicated, the same numbers in different drawings represent the same or similar elements. The embodiments described in the following exemplary embodiments do not represent all embodiments consistent with the embodiments of the present invention. They are merely examples of devices and methods consistent with some aspects of the embodiments of the present invention as detailed in the appended claims.

[0069] It will be understood that the terms "first," "second," and the like used in the present invention may be used herein to describe various concepts, but unless otherwise specified, these concepts are not limited by these terms. These terms are merely used to distinguish one concept from another. For example, without departing from the scope of the embodiments of the present invention, first information may also be referred to as second information, and similarly, second information may also be referred to as first information. Depending on the context, the words "if" and "if" as used herein may be interpreted as "at the time of," "when," or "in response to a determination."

[0070] The terms "at least one", "plurality", "each", "any", etc. used in the present invention include at least one, two or more, multiple, two or more, each refers to each of the corresponding multiple, and any refers to any one of the multiple.

[0071] Unless otherwise defined, all technical and scientific terms used herein have the same meaning as commonly understood by those skilled in the art to which the present invention pertains. The terms used herein are for the purpose of describing embodiments of the present invention only and are not intended to limit the present invention.

[0072] The green peacock ecological monitoring method based on singing provided by the embodiment of the present invention relates to the field of acoustic signal processing technology. The green peacock ecological monitoring method based on singing provided by the embodiment of the present invention can be applied to a terminal, can also be applied to a server, and can also be software running in a terminal or a server. In some embodiments, the terminal can be a smart phone, a tablet computer, a laptop computer, a desktop computer, a smart speaker, a smart watch, and a car terminal, etc., but is not limited to this; the server side can be configured as an independent physical server, or as a server cluster or distributed system composed of multiple physical servers, and can also be configured as a cloud server that provides basic cloud computing services such as cloud services, cloud databases, cloud computing, cloud functions, cloud storage, network services, cloud communications, middleware services, domain name services, security services, CDN, and big data and artificial intelligence platforms. The server can also be a node server in a blockchain network; the software can be an application that implements the green peacock ecological monitoring method based on singing, etc., but is not limited to the above forms.

[0073] The present invention can be used in a wide variety of general or special computer system environments or configurations. For example: personal computers, server computers, handheld or portable devices, tablet devices, multiprocessor systems, microprocessor-based systems, set-top boxes, programmable consumer electronics, network PCs, minicomputers, mainframe computers, distributed computing environments including any of the above systems or devices, and the like. The present invention can be described in the general context of computer-executable instructions executed by a computer, such as program modules. Generally, program modules include routines, programs, objects, components, data structures, etc. that perform specific tasks or implement specific abstract data types. The present invention can also be practiced in distributed computing environments in which tasks are performed by remote processing devices connected via a communications network. In a distributed computing environment, program modules can be located in local and remote computer storage media, including storage devices.

[0074] like Figure 1 FIG. 1 is a schematic diagram of an implementation environment provided by an embodiment of the invention. Figure 1 , the implementation environment includes at least one terminal 102 and a server 101. The terminal 102 and the server 101 can be connected to the network in a wireless or wired manner to complete data transmission and exchange.

[0075] Server 101 can be an independent physical server, or a server cluster or distributed system composed of multiple physical servers. It can also be a cloud server that provides basic cloud computing services such as cloud services, cloud databases, cloud computing, cloud functions, cloud storage, network services, cloud communications, middleware services, domain name services, security services, CDN (Content Delivery Network), as well as big data and artificial intelligence platforms.

[0076] In addition, server 101 can also be a node server in a blockchain network. Blockchain is a new application model of computer technologies such as distributed data storage, peer-to-peer transmission, consensus mechanism, and encryption algorithm.

[0077] Terminal 102 may be a smartphone, tablet computer, laptop computer, desktop computer, smart speaker, smartwatch, etc. Terminal 102 may also be a vehicle-mounted terminal of the various device types described above, but is not limited thereto. Terminal 102 and server 101 may be connected directly or indirectly via wired or wireless communication, which is not limited in this embodiment of the present invention.

[0078] Based on the example Figure 1 In the implementation environment shown, an embodiment of the present invention provides a green peacock ecological monitoring method based on singing. The following is explained using the green peacock ecological monitoring method based on singing applied to the server 101 as an example. It can be understood that the method can also be applied to the terminal 102.

[0079] Reference Figure 2 , Figure 2 The flowchart of the green peacock ecology monitoring method based on the sound provided by the embodiment of the present invention is applied to the server. The execution subject of the method can be any of the aforementioned computer devices (including servers or terminals). Figure 2 , the method may include the following steps:

[0080] Constructing a target data set; the target data set includes a green peacock sound data set, a bird sound data set, an interference sound data set, and an environmental sound data set;

[0081] Constructing a deep learning network model based on the target data set; the deep learning network model includes a sound event detection model, a bird sound recognition model, and a green peacock sound recognition model;

[0082] Collect sound data from the target monitoring area to obtain target sound data;

[0083] Extracting sound information from the target sound data to obtain target information, and identifying the target information using the deep learning network model to obtain a sound recognition result; the sound recognition result includes the sound of a green peacock, various bird calls, interference sounds, and environmental sounds;

[0084] Green peacock ecological data statistics are performed based on the sound recognition results.

[0085] In some embodiments, constructing the target dataset comprises the following steps:

[0086] Deploy a monophonic recording device in areas where green peacocks are likely to appear. The microphone of the monophonic recording device is omnidirectional, has an operating frequency band of 200 Hz-16,000 Hz, a sampling rate of 32,000 Hz or higher, and a sampling accuracy of 16 bits.

[0087] Based on the audio data collected by the monophonic recording device, the green peacock's call type is identified and the behavioral information corresponding to the call is annotated to generate label data for each type of green peacock's call;

[0088] Based on the public information on the public platform, determine the list of bird species in the monitored area and obtain the song data of these bird species;

[0089] Screening out various types of interference sounds based on the audio data collected by the monophonic recording device, including: car driving sounds, car horns, dog barking, human voices, and construction sounds, and annotating the screened interference sound data to generate interference sound label data;

[0090] Download rain, wind, and thunder sound data from open source websites to construct an environmental sound dataset;

[0091] The target data set is obtained by summarizing the label data of each type of green peacock sound, the sound data of bird species, the label data of interference sounds, and the environmental sound data set.

[0092] In some embodiments, constructing a deep learning network model based on the target data set includes the following steps:

[0093] All types of green peacock calls are considered as an acoustic event. Each type of environmental sound is considered as an acoustic event. Each type of interference sound is considered as an acoustic event. Some bird sound data is selected and combined together to form a bird sound event. The pre-trained logmelBase-Frame-Level model is used as the initial sound event detection model. The collected acoustic event detection dataset is used to retrain the logmelBase-Frame-Level model to obtain the final sound event detection model.

[0094] Each type of green peacock song dataset and various types of bird song datasets were sliced into 2s slices and randomly divided into training set, validation set and test set in a ratio of 7:2:1. The bird sound recognition model and the green peacock song recognition model were trained.

[0095] In some embodiments, retraining the logmelBase-Frame-Level model with the collected acoustic event detection dataset to obtain a final sound event detection model comprises the following steps:

[0096] In data preprocessing, the input signal is first subjected to Fourier transform to obtain the spectrum of the input signal. Then, the power spectrum density is calculated and the power spectrum is processed through a Mel filter. The transfer function of the Mel filter is defined as:

[0097]

[0098] Where H(mk) represents the frequency response function of the Mel filter; F center[m] and F width[m] are the center frequency and width of the mth filter respectively;

[0099] Take the logarithm of the Mel filter processing result to get V[m], V[m] represents the output of the m-th Mel filter group, and serialize the processed Mel filter output V[m] into a feature vector, which is used as the input of the logmelBase-Frame-Level model;

[0100] Based on the features extracted by the Log-Mel process, an embedded representation at the segment level or frame level is obtained;

[0101] The embedding representation of the query set is multiplied by the initialized weight matrix, and the prediction result is obtained through the softmax function, thereby obtaining a sound event detection model. The prediction output includes four types of sound event detection results: green peacock sound, environmental sound, interference sound, and bird sound.

[0102] Among them, the network structure of the sound event detection model includes 2 BasicBlock layers, 2 CNN layers and 1 linear layer.

[0103] In some embodiments, the data sets of each type of green peacock song and various types of bird song data sets are sliced into 2s slices and randomly divided into training set, validation set and test set in a ratio of 7:2:1, and the bird sound recognition model and the green peacock song recognition model are trained, including the following steps:

[0104] The data set is filtered using a bandpass filtering method to obtain a signal y[n] after filtering; wherein, the filtering adopts a window function method to implement a bandpass filter, and the window function adopts a Hanning window;

[0105] Perform Chirplet transformation on the filtered signal to obtain a Chirplet speech graph;

[0106] Among them, Chirplet transformation first requires defining Chirplet atoms. A Chirplet atom can be expressed as: Where f0 represents the center frequency; α represents the frequency modulation rate; t0 represents the center time of the atom; represents a window function, which is 1 from time t0-Δ to t0+α, otherwise it is 0; ψ[n] represents a Chirplet atomic operator;

[0107] After the Chirplet atom library is constructed, for each atom in the y[n] sequence obtained through filtering and the atom library Calculate discrete Chirplet transform coefficients The calculation expression is: in, is the complex conjugation of the atom;

[0108] The Chirplet graph is constructed by taking the modulus of the calculated Chirplet transformation coefficients;

[0109] The obtained Chirplet spectrogram is input into the VGGish model for feature extraction. After feature extraction in the VGGish model, the feature data is input into the XGBoost algorithm for classification, and the corresponding probability of each type of green peacock call and the probability of each bird species are output.

[0110] In some embodiments, performing green peacock ecological data statistics based on the sound recognition results includes the following steps:

[0111] The green peacock's calls were counted over time by monitoring area, and the frequency of calls at different times of the day, as well as seasonal variations, were analyzed to generate rhythmic diagrams showing the green peacock's call activity patterns in each monitoring area.

[0112] The various types of green peacock calls are counted by time, and the collected sound data and corresponding labels will be subjected to time series analysis. The frequency of occurrence of each sound type will be counted and associated with time to determine the frequency of green peacock calls in specific time periods. The time when human interference sounds occur will also be counted to analyze the impact of interference sounds on green peacock activities.

[0113] In some embodiments, the performing of green peacock ecological data statistics based on the sound recognition results further includes the following steps:

[0114] Statistical analysis was used to evaluate the relationship between the intensity and frequency of interfering sounds and the singing patterns of green peacocks, and to analyze the effects of interfering sounds on the singing behavior and daily activities of green peacocks.

[0115] Calculate the sound pressure level and spectral components of human interference sounds, and analyze the impact of the interference sound size and frequency components on the behavior of green peacocks;

[0116] The potential effects of sound intensity on the behavior of green peacocks were assessed by analyzing the sound pressure levels of audio clips measured by a sound level meter;

[0117] The frequency distribution of the sound was measured by a spectrum analyzer to analyze the effects of sounds in different frequency ranges on the behavior of the green peacock;

[0118] Observe and count other bird species coexisting in the monitoring area, and analyze the interactions between coexisting bird species.

[0119] The following describes in detail the implementation process of the present invention in specific application scenarios with reference to the accompanying drawings:

[0120] A method for monitoring the ecology of green peacocks based on their singing, comprising the following steps:

[0121] S1. Build a dataset;

[0122] S2. Build deep learning network models (sound event detection, bird sound recognition, and green peacock call recognition);

[0123] S3, sound data collection in the monitoring area;

[0124] S4, sound data information extraction;

[0125] S5. Green peacock ecological data statistics.

[0126] 1. The specific details of S1 are:

[0127] 1) Recording Equipment Performance: Deploy monophonic recording equipment in areas where green peacocks are likely to appear. The monophonic recording equipment should have an omnidirectional microphone, an operating frequency band of 200 Hz to 16,000 Hz, a sampling rate of at least 32,000 Hz, and a sampling accuracy of 16 bits.

[0128] 2) Construction of a Green Peafowl Call Dataset. The audio output from a monophonic recording device was manually identified by identifying the type of green peacock call and annotating the behavioral information corresponding to the call, generating labeled data for each green peacock call type. For call types with fewer than 100, monophonic recording devices were deployed in the zoo to collect data on these call types, ensuring that there are at least 100 calls for each type. The audio files were formatted using lossless compression (WAV).

[0129] 3) Constructing a bird sound dataset. We determined the bird species list within the monitored area through information published by government agencies, academic literature, and the China Birdwatching Recording Center. We downloaded the call data for these bird species from public databases and open-source websites, using a unified sampling frequency of 32kHz and a quantization accuracy of 16 bits. The constructed bird sound dataset contained at least 100 data points for each bird sound. Each data point was saved as an audio file using lossless compression in the WAV format.

[0130] 4) Construction of interference sound dataset. The audio files collected by the recording equipment are manually screened to remove various types of interference sounds, including: car driving sounds, car horns, dog barking, human voices, and construction sounds. The manually screened interference sound data is annotated to generate interference sound label data. For interference sounds with less than 100 data, they are downloaded and supplemented from the open source sound website, so that each type of interference sound data is no less than 100. Each data is saved as an audio file, and the audio file format uses lossless compression wav format.

[0131] 5) Construction of an ambient sound dataset. Rain, wind, and thunder sound data were downloaded from open-source websites, with a unified sampling frequency of 32kHz and a quantization precision of 16 bits. When constructing the ambient sound dataset, we ensured that there were at least 100 data points for each type of ambient sound. If there were fewer than 100, we collected and labeled them using a single-channel recording device. Each data point was saved as an audio file using lossless compression in the WAV format.

[0132] 2. The specific details of S2 are:

[0133] 1) Sound event detection. All types of green peacock calls are considered as an acoustic event; each type of environmental sound is considered as an acoustic event; each type of interference sound is considered as an acoustic event; and some of the various bird sound data are selected and combined together as a bird sound event. The sound event detection model uses the logmelBase-Frame-Level model and is pre-trained using an existing open source dataset. The collected acoustic event detection dataset is then used to retrain the logmelBase-Frame-Level model. The framework of the logmelBase-Frame-Level model is as follows: Figure 3 shown.

[0134] In data preprocessing, the model first performs Fourier transform on the input signal to obtain its spectrum, then calculates the power spectrum density and passes the power spectrum through the Mel filter. The transfer function of the Mel filter is defined as:

[0135]

[0136] Here, Fcenter[m] and Fwidth[m] are the center frequency and width of the mth filter, respectively. The logarithm of the result after passing through the Mel filter is taken to obtain V[m], simulating the human ear's logarithmic perception of sound intensity. Here, V[m] represents the output of the mth Mel filter bank. Finally, the processed Mel filter output V[m] is serialized into a feature vector, which serves as the input to the logmelBase-Frame-Level model. This serialized feature vector is the final output of the Log-Mel feature.

[0137] Through the Log-Mel feature extraction process above, the extracted features are input into the network to obtain a segment-level (or frame-level) embedding representation. Finally, the embedding representation of the query set is multiplied by the initialized weight matrix and a softmax function is applied to obtain the prediction result. This prediction outputs four types of sound event detection results: green peacock sound, environmental sound, interference sound, and bird sound.

[0138] The model's network structure includes two BasicBlock layers, two CNN layers, and one linear layer. BasicBlock is the core building block in ResNet (residual network). It directly adds input to output through skip connections, helping to alleviate the vanishing gradient problem and accelerate network convergence.

[0139] 2) Green Peafowl and Birdsong Recognition Network. Each Green Peafowl song type dataset and various bird song types were sliced into 2-second slices and randomly divided into training, validation, and test sets in a ratio of 7:2:1. Research has shown that the frequency of the songs of Green Peafowl and other birds with similar habits is concentrated between 0.4kHz and 16kHz. Bandpass filtering was used to further improve the recognition accuracy of the model. Filtering was implemented using a Hanning window function to achieve a bandpass filter. After filtering, the signal y[n] was obtained.

[0140] refer to Figure 4 , the filtered signal is subjected to Chirplet transformation to obtain a Chirplet graph. Chirplet transformation first requires the definition of Chirplet atoms. A Chirplet atom can be expressed as:

[0141]

[0142] Where f0 is the center frequency, α is the frequency modulation rate (chirprate), and t0 is the center time of the atom. Is a window function, usually a rectangular window, which is 1 in the time range t0-Δ to t0+Δ, and 0 otherwise. A series of parameters with different values are selected to construct the atom library, where these parameters include f0, α and t0. After the Chirplet atom library is constructed, for each atom in the y[n] sequence obtained after filtering and the atom library Compute discrete Chirplet transform coefficients:

[0143]

[0144] in, The Chirplet graph is constructed by taking the calculated Chirplet transform coefficients modulo the complex conjugate of the atom.

[0145] The obtained Chirplet spectrogram is input into the VGGish model for feature extraction. After the VGGish model performs feature extraction, the feature data is input into the XGBoost algorithm for classification. The corresponding probability of each type of green peacock song and the probability of each bird species are output. The overall process of green peacock and bird song recognition model training is as follows: Figure 5 shown.

[0146] 3. The specific details of S3 are:

[0147] 1) Deployment of recording equipment. A mono recording device is deployed in the green peacock monitoring area. The mono recording microphone is omnidirectional and has an operating frequency band range of 200Hz-16000Hz. The sampling rate of the recording device is set to above 32000Hz, and the sampling accuracy is 16 bits. In order to prevent human factors from affecting the activities of the green peacock, the recording device supports 4G wireless network data transmission and can be powered by a variety of methods (solar energy, 220V AC power, battery). The recording equipment is installed in the green peacock monitoring area, including streams, woodlands and woodland boundaries, with the microphone 1.5 meters to 3 meters above the ground.

[0148] 2) Recording device data storage and transmission. The recording device operates via sound event triggering. When a sound signal is detected, it records the recorded data. The recorded data is transmitted via 4G wireless to a remote server and stored on an SD card for backup. Other metadata associated with the recorded data (geographic location, recording time, temperature, and humidity) is also transmitted and stored.

[0149] 4. The specific details of S4 are as follows:

[0150] 1) The recording data sent back by the recording device is processed on the remote server. Sound event detection is performed on the recording data using the trained sound event detection model, including the singing of green peacocks, bird sounds, interference sounds (including car sounds, car horns, dog barking, human voices, construction sounds, bird sounds), and environmental sounds (rain, wind, thunder).

[0151] 2) Save the detected green peacock sounds of various types, bird sounds, interference sounds (including car driving sounds, car horns, dog barking, human voices, construction sounds), environmental sound data fragments, and the corresponding time, geographical location of the recording equipment, temperature, humidity and other data.

[0152] 3) Green peacock sound recognition: The data corresponding to the detected green peacock sound events are re-recognized using the green peacock sound recognition network model. If the confidence level exceeds 0.8, the green peacock sound data and its sound type information are saved.

[0153] 4) Birdsong Recognition: The detected bird song segments are used to identify the specific species through the bird song recognition network model. If the confidence level exceeds 0.8, the species song data and species information are saved.

[0154] 5. The specific details of S5 are:

[0155] 1) Analyze the frequency of green peacock calls by time of day, analyzing the frequency of calls at different times of the day and any seasonal variations. Using this data, the present invention can create detailed rhythmic diagrams to more intuitively demonstrate the patterns of green peacock call activity in each monitoring area.

[0156] 2) Counting the various types of green peacock calls by time. The collected sound data and corresponding labels are subjected to a time series analysis, counting the frequency of each sound type and correlating it with time. This method allows the present embodiment to determine whether green peacock calls are more frequent during specific time periods. The timing of occurrence of various human interference sounds, such as dog barking, construction, and cars, is also counted to analyze their impact on green peacock activity.

[0157] 3) In addition, embodiments of the present invention will also use statistical analysis methods, such as correlation analysis or regression models, to evaluate the relationship between the intensity and frequency of the interfering sounds and the green peacock's singing patterns. This will help embodiments of the present invention understand whether the interfering sounds change the green peacock's singing behavior and whether they affect the green peacock's daily activities.

[0158] 4) Calculate the sound pressure level and spectral components of various human interference sounds such as dog barking / construction / cars, and analyze the impact of the interference sound size and frequency components on the behavior of the green peacock. By performing sound pressure level analysis on the audio clips measured by the sound level meter, the embodiment of the present invention can evaluate the potential impact of sound intensity on the behavior of the green peacock. Sound pressure level is a physical quantity that measures the energy of sound, which reflects the amplitude of the sound wave. In animal behavior research, changes in sound pressure level may have a significant impact on the communication, foraging and reproductive behavior of animals. For example, high-decibel noise may interfere with the call of the green peacock, affect its reproductive success rate, or cause a stress response, thereby affecting its daily activity pattern.

[0159] Furthermore, by measuring the frequency distribution of sounds using a spectrum analyzer, the present invention provides a more detailed understanding of the impact of sounds of different frequency ranges on the behavior of green peacocks. The frequency distribution of sounds can help the present invention identify the main frequency components that may affect green peacock behavior. For example, sounds of certain frequencies may stimulate a green peacock's alert behavior, while sounds of other frequencies may interfere with its foraging or resting behavior.

[0160] By analyzing the above data, researchers can assess the specific effects of specific frequencies and sound pressure levels on the behavior of green peacocks and propose corresponding protection measures.

[0161] 5) Statistics of coexisting bird species and activity rhythms. While conducting the behavioral study of the green peacock, the embodiment of the present invention will also observe and count other bird species coexisting in the monitoring area. This includes recording their activity rhythms, such as active time of day, seasonal activity changes, etc. Through time series analysis, the embodiment of the present invention can reveal the interactions between coexisting bird species. For example, some species may choose to rest when the green peacock is active to avoid competition. This will help the embodiment of the present invention understand whether different bird species share similar ecological niches, or whether they have developed different behavioral strategies to share the same living environment. By arranging the above information, a data statistical table is constructed as shown in Table 1:

[0162] Table 1 Data statistics

[0163]

[0164] Another aspect of the present invention provides a green peacock ecology monitoring system based on singing, comprising:

[0165] The first module is used to construct a target data set; the target data set includes a green peacock sound data set, a bird sound data set, an interference sound data set, and an environmental sound data set;

[0166] The second module is used to build a deep learning network model based on the target data set; the deep learning network model includes a sound event detection model, a bird sound recognition model, and a green peacock sound recognition model;

[0167] The third module is used to collect sound data from the target monitoring area to obtain target sound data;

[0168] A fourth module is configured to extract sound information from the target sound data to obtain target information, and identify the target information using the deep learning network model to obtain a sound recognition result; the sound recognition result includes the sound of a green peacock, various bird calls, interference sounds, and environmental sounds;

[0169] The fifth module is used to collect ecological data of green peacocks based on the sound recognition results.

[0170] It can be understood that the contents of the above method embodiments are all applicable to the present system embodiments, the functions specifically implemented by the present system embodiments are the same as those of the above method embodiments, and the beneficial effects achieved are also the same as those achieved by the above method embodiments.

[0171] An embodiment of the present invention further provides an electronic device comprising a memory and a processor. The memory stores a computer program, and the processor executes the computer program to implement the aforementioned method for monitoring the ecology of green peacocks based on their chirping. The electronic device can be any intelligent terminal, including a tablet computer and an in-vehicle computer.

[0172] It can be understood that the contents of the above method embodiments are applicable to the present device embodiments, the functions specifically implemented by the present device embodiments are the same as those of the above method embodiments, and the beneficial effects achieved are also the same as those achieved by the above method embodiments.

[0173] An embodiment of the present invention also provides a computer-readable storage medium, which stores a computer program. When the computer program is executed by a processor, it implements the above-mentioned green peacock ecological monitoring method based on singing.

[0174] It can be understood that the contents of the above method embodiments are all applicable to the present storage medium embodiment, the functions specifically implemented by the present storage medium embodiment are the same as those of the above method embodiments, and the beneficial effects achieved are also the same as those achieved by the above method embodiments.

[0175] The memory, as a non-transient computer-readable storage medium, can be used to store non-transient software programs and non-transient computer executable programs. In addition, the memory may include a high-speed random access memory and may also include a non-transient memory, such as at least one disk storage device, a flash memory device, or other non-transient solid-state storage device. In some embodiments, the memory may optionally include a memory remotely arranged relative to the processor, and these remote memories may be connected to the processor via a network. Examples of the above-mentioned network include, but are not limited to, the Internet, an intranet, a local area network, a mobile communication network, and combinations thereof.

[0176] It should be noted that in various specific embodiments of the present invention, when it comes to the need to perform relevant processing based on data related to the user's identity or characteristics, such as user information, user behavior data, user historical data, and user location information, the user's permission or consent will be obtained first, and the collection, use, and processing of such data will comply with relevant laws, regulations, and standards. In addition, when the embodiment of the present invention needs to obtain the user's sensitive personal information, it will obtain the user's separate permission or consent through a pop-up window or jump to a confirmation page. After clearly obtaining the user's separate permission or consent, the necessary user-related data for the normal operation of the embodiment of the present invention will be obtained.

[0177] In some optional embodiments, the function / operation mentioned in the block diagram may not occur in the order mentioned in the operation diagram. For example, depending on the function / operation involved, the two boxes shown in succession can actually be executed substantially simultaneously or the boxes can sometimes be executed in reverse order. In addition, the embodiment presented and described in the flow chart of the present invention is provided in an exemplary manner for the purpose of providing a more comprehensive understanding of the technology. The disclosed method is not limited to the operation and logic flow presented herein. Optional embodiments are contemplated in which the order of the various operations is changed and the sub-operations described as a part of a larger operation are performed independently.

[0178] Furthermore, although the present invention is described in the context of functional modules, it should be understood that, unless otherwise indicated, one or more of the functions and / or features described may be integrated into a single physical device and / or software module, or one or more functions and / or features may be implemented in separate physical devices or software modules. It will also be understood that a detailed discussion of the actual implementation of each module is not necessary for understanding the present invention. More specifically, given the properties, functions, and internal relationships of the various functional modules in the devices disclosed herein, the actual implementation of the module will be understood within the ordinary skill of an engineer. Therefore, a person skilled in the art using ordinary skill will be able to implement the present invention set forth in the claims without undue experimentation. It will also be understood that the specific concepts disclosed are merely illustrative and are not intended to limit the scope of the present invention, which is determined by the full scope of the appended claims and their equivalents.

[0179] If the functions are implemented in the form of software functional units and sold or used as independent products, they can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present invention, or the part that contributes to the prior art, or the part of the technical solution, can be embodied in the form of a software product. The computer software product is stored in a storage medium and includes several instructions for enabling a computer device (which can be a personal computer, a server, or a network device, etc.) to execute all or part of the steps of the method described in each embodiment of the present invention. The aforementioned storage medium includes various media that can store program codes, such as a USB flash drive, a mobile hard disk, a read-only memory (ROM), a random access memory (RAM), a magnetic disk, or an optical disk.

[0180] The logic and / or steps represented in the flowcharts or otherwise described herein, for example, can be considered as an ordered list of executable instructions for implementing the logical functions, and can be embodied in any computer-readable medium for use by, or in conjunction with, an instruction execution system, apparatus, or device (e.g., a computer-based system, a system including a processor, or other system that can fetch and execute instructions from an instruction execution system, apparatus, or device). For purposes of this specification, a "computer-readable medium" can be any device that can contain, store, communicate, propagate, or transport a program for use by, or in conjunction with, an instruction execution system, apparatus, or device.

[0181] More specific examples (a non-exhaustive list) of computer-readable media include the following: an electrical connection with one or more wires (electronic devices), a portable computer disk cartridge (magnetic devices), a random access memory (RAM), a read-only memory (ROM), an erasable and programmable read-only memory (EPROM or flash memory), a fiber optic device, and a portable compact disc read-only memory (CDROM). In addition, the computer-readable medium may even be paper or other suitable medium on which the program is printed, since the program may be obtained electronically, for example, by optically scanning the paper or other medium, followed by editing, deciphering, or processing in another suitable manner as necessary, and then stored in a computer memory.

[0182] It should be understood that various parts of the present invention can be implemented using hardware, software, firmware, or a combination thereof. In the above-described embodiments, multiple steps or methods can be implemented using software or firmware stored in a memory and executed by a suitable instruction execution system. For example, if implemented using hardware, as in another embodiment, any one of the following technologies known in the art or a combination thereof can be used: a discrete logic circuit having a logic gate circuit for implementing a logic function on a data signal, an application-specific integrated circuit having a suitable combination of logic gate circuits, a programmable gate array (PGA), a field programmable gate array (FPGA), etc.

[0183] Throughout this specification, reference to terms such as "one embodiment," "some embodiments," "examples," "specific examples," or "some examples" means that a specific feature, structure, material, or characteristic described in conjunction with that embodiment or example is included in at least one embodiment or example of the present invention. In this specification, schematic representations of the above terms do not necessarily refer to the same embodiment or example. Furthermore, the specific features, structures, materials, or characteristics described may be combined in any suitable manner in any one or more embodiments or examples.

[0184] While embodiments of the present invention have been shown and described, it will be appreciated by those skilled in the art that various changes, modifications, substitutions, and variations may be made to the embodiments without departing from the principles and spirit of the invention, and that the scope of the invention is defined by the claims and their equivalents.

[0185] The above is a specific description of the preferred implementation of the present invention, but the present invention is not limited to the embodiments. Those skilled in the art can make various equivalent modifications or substitutions without violating the spirit of the present invention. These equivalent modifications or substitutions are all included in the scope defined by the claims of the present invention.

Claims

1. A method for monitoring the ecology of green peacocks based on their singing, characterized in that: The following steps are involved: Build the target dataset; The target data sets include green peacock sound data sets, bird sound data sets, interference sound data sets, and environmental sound data sets; Constructing a deep learning network model based on the target data set; the deep learning network model includes a sound event detection model, a bird sound recognition model, and a green peacock sound recognition model; Collect sound data from the target monitoring area to obtain target sound data; Extracting sound information from the target sound data to obtain target information, and identifying the target information using the deep learning network model to obtain a sound recognition result; the sound recognition result includes the sound of a green peacock, various bird calls, interference sounds, and environmental sounds; Conducting green peacock ecological data statistics based on the sound recognition results; The method of performing green peacock ecological data statistics according to the sound recognition result comprises the following steps: The green peacock's calls were counted over time by monitoring area, and the frequency of calls at different times of the day and seasonal variations were analyzed to generate a rhythmic diagram showing the green peacock's call activity patterns in each monitoring area. The various types of green peacock calls will be counted over time. The collected sound data and corresponding labels will be subjected to time series analysis. The frequency of each sound type will be counted and associated with time to determine the frequency of green peacock calls in specific time periods. The time when human interference sounds occur will also be counted to analyze the impact of interference sounds on green peacock activities. Statistical analysis was used to evaluate the relationship between the intensity and frequency of interfering sounds and the singing patterns of green peacocks, and to analyze the effects of interfering sounds on the singing behavior and daily activities of green peacocks. Calculate the sound pressure level and spectral components of human interference sounds, and analyze the impact of the interference sound size and frequency components on the behavior of green peacocks; The potential effects of sound intensity on the behavior of green peacocks were assessed by analyzing the sound pressure levels of audio clips measured by a sound level meter; The frequency distribution of the sound was measured by a spectrum analyzer to analyze the effects of sounds in different frequency ranges on the behavior of the green peacock; Observe and count other bird species coexisting in the monitoring area, and analyze the interactions between coexisting bird species.

2. The method for monitoring green peacock ecology based on sound according to claim 1, characterized in that: The target dataset is constructed, comprising the following steps: Deploy a monophonic recording device in areas where green peacocks are likely to appear. The monophonic recording device has an omnidirectional microphone, an operating frequency band of 200 Hz-16,000 Hz, a sampling rate of 32,000 Hz or higher, and a sampling accuracy of 16 bits. Based on the audio data collected by the monophonic recording device, the green peacock's call type is identified and the behavioral information corresponding to the call is annotated to generate label data for each type of green peacock's call; Based on the public information on the public platform, determine the list of bird species in the monitored area and obtain the song data of these bird species; Screening out various types of interference sounds based on the audio data collected by the monophonic recording device, including: car driving sounds, car horns, dog barking, human voices, and construction sounds, and annotating the screened interference sound data to generate label data of the interference sounds; Download rain, wind, and thunder sound data from open source websites to construct an environmental sound dataset; The target data set is obtained by summarizing the label data of each type of green peacock sound, the sound data of bird species, the label data of interference sounds, and the environmental sound data set.

3. The method for monitoring green peacock ecology based on sound according to claim 1, characterized in that: The method of constructing a deep learning network model according to the target data set includes the following steps: All types of green peacock calls are considered as an acoustic event. Each type of environmental sound is considered as an acoustic event. Each type of interference sound is considered as an acoustic event. Some bird sound data is selected and combined together to form a bird sound event. The pre-trained logmelBase-Frame-Level model is used as the initial sound event detection model. The collected acoustic event detection dataset is used to retrain the logmelBase-Frame-Level model to obtain the final sound event detection model. Each type of green peacock song dataset and various types of bird song datasets were sliced into 2s slices and randomly divided into training set, validation set and test set in a ratio of 7:2:

1. The bird sound recognition model and the green peacock song recognition model were trained.

4. The method for monitoring green peacock ecology based on sound according to claim 3, characterized in that: The method of retraining the logmelBase-Frame-Level model with the collected acoustic event detection dataset to obtain the final sound event detection model includes the following steps: In data preprocessing, the input signal is first subjected to Fourier transform to obtain the spectrum of the input signal. Then, the power spectrum density is calculated and the power spectrum is processed through a Mel filter. The transfer function of the Mel filter is defined as: in, Represents the frequency response function of the Mel filter; and They are The center frequency and width of each filter; Take the logarithm of the Mel filter processing result to get V[m], V[m] represents the output of the m-th Mel filter group, and serialize the processed Mel filter output V[m] into a feature vector, which is used as the input of the logmelBase-Frame-Level model; Based on the features extracted by the Log-Mel process, an embedded representation at the segment level or frame level is obtained; The embedding representation of the query set is multiplied by the initialized weight matrix, and the prediction result is obtained through the softmax function, thereby obtaining a sound event detection model. The prediction output includes four types of sound event detection results: green peacock sound, environmental sound, interference sound, and bird sound. Among them, the network structure of the sound event detection model includes 2 BasicBlock layers, 2 CNN layers and 1 linear layer.

5. The method for monitoring green peacock ecology based on sound according to claim 3, characterized in that: The data sets of each type of green peacock song and various types of bird song data sets are sliced into 2s slices and randomly divided into a training set, a validation set, and a test set in a ratio of 7:2:

1. The bird sound recognition model and the green peacock song recognition model are trained, including the following steps: The data set is filtered using a bandpass filtering method to obtain a signal y[n] after filtering; wherein, the filtering adopts a window function method to implement a bandpass filter, and the window function adopts a Hanning window; Perform Chirplet transformation on the filtered signal to obtain a Chirplet speech graph; Among them, Chirplet transformation first requires defining Chirplet atoms. A Chirplet atom is represented as: ,in, represents the center frequency; represents the frequency modulation rate; represents the central time of the atom; Represents a window function, which is in time arrive 1 if it is inside, 0 otherwise; Represents a Chirplet atomic operator; After the Chirplet atom library is constructed, for each atom in the y[n] sequence obtained through filtering and the atom library Calculate discrete Chirplet transform coefficients , the calculation expression is: ,in, is the complex conjugation of the atom; The Chirplet graph is constructed by taking the modulus of the calculated Chirplet transformation coefficients; The obtained Chirplet spectrogram is input into the VGGish model for feature extraction. After feature extraction in the VGGish model, the feature data is input into the XGBoost algorithm for classification, and the corresponding probability of each type of green peacock call and the probability of each bird species are output.

6. A system for implementing the green peacock ecology monitoring method based on singing as claimed in any one of claims 1 to 5, characterized in that: include: The first module is used to construct the target dataset; The target data sets include green peacock sound data sets, bird sound data sets, interference sound data sets, and environmental sound data sets; The second module is used to build a deep learning network model based on the target data set; the deep learning network model includes a sound event detection model, a bird sound recognition model, and a green peacock sound recognition model; The third module is used to collect sound data from the target monitoring area to obtain target sound data; A fourth module is configured to extract sound information from the target sound data to obtain target information, and identify the target information using the deep learning network model to obtain a sound recognition result; the sound recognition result includes the sound of a green peacock, various bird calls, interference sounds, and environmental sounds; The fifth module is used to collect ecological data of green peacocks based on the sound recognition results.

7. An electronic device, characterized in that: including a processor and a memory; The memory is used to store programs; The processor executes the program to implement the method according to any one of claims 1 to 5.

8. A computer-readable storage medium, characterized in that The storage medium stores a program, and the program is executed by a processor to implement the method according to any one of claims 1 to 5.

Images