Green low-carbon coastal wetland intelligent bird watching method and system
By using multimodal sensor networks and local intelligent processing technology, the problems of low automation and limited data in coastal wetland ecological monitoring have been solved, enabling efficient and accurate waterbird monitoring and ecological status perception, and improving the credibility of scientific research results and public participation.
Patent Information
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2026-01-06
- Publication Date
- 2026-05-26
Smart Images
Figure CN122087556A_ABST
Abstract
Description
Technical Field
[0001] This application belongs to the field of ecological monitoring, and specifically relates to a green, low-carbon, intelligent birdwatching method for coastal wetlands. Background Technology
[0002] With the increasing impact of global climate change and human activities, coastal wetland ecosystems and their wintering waterbirds are facing increasingly severe survival pressures. In current wetland conservation and birdwatching practices, the relevant infrastructure generally adopts the form of traditional wooden observation towers or simple observation stations. Their core design focuses on providing observers with basic visual concealment and physical protection, lacking the ability to automatically collect environmental data and biological signals.
[0003] Specifically, existing technical solutions require extensive manual intervention in species identification, population statistics, and behavioral recording, resulting in low efficiency and susceptibility to subjective factors. This makes it difficult to achieve long-term, continuous, and standardized scientific monitoring. Furthermore, data collection is often limited to a single dimension, focusing only on image and video recordings and lacking the simultaneous acquisition and cross-validation of multimodal information such as bird sounds and environmental factors. This restricts the depth and reliability of ecological pattern analysis. In addition, the layout of facilities and the observation activities themselves do not adequately consider the sensitivity of wetland ecosystems. Frequent personnel entry and exit, equipment noise, and light interference may disturb waterbird habitats, violating the primary principle of ecological protection.
[0004] Therefore, there is an urgent need for an integrated smart birdwatching technology solution for coastal wetlands that can achieve intelligent sensing, low-carbon operation, data fusion, and take into account the needs of scientific research monitoring and public education. Summary of the Invention
[0005] To address the aforementioned issues, this application provides a green and low-carbon smart birdwatching method for coastal wetlands, which prioritizes improving the efficiency, continuity, and objectivity of monitoring work.
[0006] A green, low-carbon, and intelligent birdwatching method for coastal wetlands includes the following steps: Based on the environmental parameter data of the target coastal wetland and waterbird activity information, observation points were selected; At the observation points, monitoring data, including at least waterbird images, waterbird sounds, and wetland environmental parameters, are collected based on a multimodal sensor network. The monitoring data is subjected to local multimodal data fusion and identification processing to obtain waterbird species identification results. The local multimodal data fusion and identification processing includes the fusion analysis and identification of waterbird images and waterbird sounds. Based on the waterbird species identification results, ecological observation information services are provided to the public. The selection of observation points includes: Based on pre-monitoring data, areas where waterbirds stay for longer than a preset time are identified as high-frequency activity areas for waterbirds. Determine the geographical location of at least one observation point within the high-frequency activity area of waterbirds and / or within a safe observation range outside the high-frequency activity area of waterbirds.
[0007] Multimodal data fusion and recognition processing includes: Denoising and normalization preprocessing were performed on waterbird images; Noise reduction, filtering preprocessing, and spectrum analysis were performed on the sounds of waterbirds to extract sound features; A convolutional neural network model was used to identify the species of waterbirds after denoising and normalization, and the image recognition results were obtained. The support vector machine model was used to classify and identify the sounds of waterbirds, and the results of the waterbird sound recognition were obtained. By correlating coastal wetland environmental parameter data with waterbird images and sounds, and obtaining waterbird species identification results, we can obtain waterbird species identification results.
[0008] The method also includes: Each event in which a waterbird species is identified based on monitoring data is associated with the fixed geographic coordinates of the corresponding observation point to generate a waterbird observation record with spatial location information and timestamp information.
[0009] The method also includes: Upload waterbird observation records with spatial location information from the observation points to the cloud platform; Based on the spatial location information in the waterbird observation records of each observation point, the cloud platform uses clustering algorithms to perform spatial distribution analysis in order to identify the gathering areas of waterbird activity.
[0010] Spatial distribution analysis includes: Based on the spatial location information in waterbird observation records, a spatial density distribution map of waterbird activity is generated using a kernel density estimation algorithm. By overlaying the spatial density distribution map with the geographic information base map of the coastal wetlands, a heat map of waterbird distribution is generated.
[0011] The method also includes: On the cloud platform, based on the timestamp information and identification results in the waterbird observation records, a time series analysis model is used to identify the behavioral patterns of waterbirds. By combining historical environmental data from observation points with waterbird observation records, a predictive model is used to construct the correlation between environmental parameters and waterbird behavior.
[0012] This application provides a green and low-carbon smart birdwatching system for coastal wetlands, including an observation point selection module, a data acquisition module, a local intelligent processing module, and an interaction module.
[0013] The observation point selection module is used to select observation points based on environmental data and bird activity information of the target coastal wetland. The data acquisition module, configured at the observation point selected by the observation point deployment module, includes a multimodal sensor network for acquiring monitoring data including at least waterbird images, waterbird sounds, and wetland environmental parameters at the observation point. The local intelligent processing module communicates with the data acquisition module to perform fusion analysis and identification of monitoring data in order to obtain waterbird species identification results. The interaction module, connected to the local intelligent processing module, is used to generate and provide ecological observation information services to the public based on the waterbird species identification results.
[0014] This application also provides an electronic device, which includes at least one processor and at least one memory, the memory being data-connected to the processor, wherein the memory stores instructions executable by at least one processor, the instructions being executed by at least one processor to enable at least one processor to perform any of the methods described above.
[0015] This application also provides a computer-storable medium storing computer instructions, which, when executed by a processor, specifically perform the steps of any of the methods described above.
[0016] This application also provides a computer program product, including computer instructions, which, when executed by a processor, specifically perform the steps of any of the methods described above.
[0017] Compared with the prior art, this application has the following advantages: Compared to existing technologies, this application scientifically selects observation points based on pre-monitoring data and integrates a multimodal sensor network to automatically collect images, sounds, and environmental data. It utilizes a local intelligent processing unit to achieve automated identification of waterbird species and behaviors. Compared to the traditional operation mode that relies on manual observation and recording, this improves the efficiency, continuity, and objectivity of monitoring work, providing a stable and reliable data foundation for long-term ecological research. At the same time, addressing the problem of single data sources in traditional observation, this application adopts multimodal data fusion identification and processing technology to simultaneously analyze the visual images and sound characteristics of waterbirds and perform correlation analysis in conjunction with real-time environmental parameters. This enables more accurate identification of waterbirds and a more comprehensive perception of their ecological status in complex wetland backgrounds, overcoming the potential misjudgments and limitations of single-dimensional data, and making research results more accurate and credible.
[0018] Other features and advantages of this application will be set forth in the description which follows, and will be apparent in part from the description, or may be learned by practicing the application. The objectives and other advantages of this application may be realized and obtained by means of the structures pointed out in the description, claims and drawings. Attached Figure Description
[0019] To more clearly illustrate the technical solutions in the embodiments of this application or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are some embodiments of this application. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.
[0020] Figure 1 A flowchart of a method according to an embodiment of this application is shown; Figure 2 A system framework diagram according to an embodiment of this application is shown; Figure 3 A schematic diagram of the structure of a smart birdwatching hut according to an embodiment of this application is shown.
[0021] In the picture: 1. Basic load-bearing module; 2. Wall enclosure module; 3. Observation window module; 4. Roof energy integration module. Detailed Implementation
[0022] To make the objectives, technical solutions, and advantages of the embodiments of this application clearer, the technical solutions of the embodiments of this application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of this application, not all embodiments. Based on the embodiments of this application, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of this application.
[0023] This invention relates to the field of ecological monitoring and environmental protection technology, and more specifically, to a green, low-carbon, intelligent birdwatching method and system for coastal wetlands.
[0024] Coastal wetlands are crucial habitats for numerous wintering waterbirds, making their ecological protection paramount. Traditional birdwatching and monitoring methods primarily rely on simple, manned observation facilities or independently operated monitoring equipment, which suffer from numerous shortcomings, including low automation, limited data dimensions, high dependence on human intervention, potential habitat disturbance, and weak public participation and experience. Existing technologies struggle to achieve efficient, accurate, low-interference, and sustainable ecological observation that also possesses scientific and educational value.
[0025] In view of this, the present invention aims to provide a green, low-carbon, intelligent birdwatching method and system for coastal wetlands to solve the above problems, specifically including the following steps: S1. Set up observation points; This step aims to determine the installation location of observation facilities based on ecological principles and environmental data, which can efficiently monitor waterbird activities while minimizing human interference, thus laying the spatial foundation for the effective operation of the entire system.
[0026] S1-1. Identify high-frequency activity areas of waterbirds based on pre-monitoring data.
[0027] Before the formal deployment of permanent facilities, several weeks of detailed pre-monitoring can be conducted. This can be achieved by using drones equipped with zoom cameras and infrared thermal imagers to perform periodic flights over the target wetlands along pre-set routes, acquiring high-resolution orthophotos, video streams, and thermal radiation data. Furthermore, infrared trigger cameras can be deployed at key ecological nodes to continuously capture ground images, aiming to obtain pre-monitoring data for the target wetlands.
[0028] Based on this pre-monitoring data, in-depth spatial and behavioral pattern analysis is conducted to guide the deployment of monitoring points. The core of the analysis lies in identifying high-frequency areas of waterbird activity and quantifying their dwell time. Specifically, methods such as stitching together UAV aerial imagery and infrared thermal imaging data, along with target detection, can be used to initially delineate the spatial range of bird flocking activities.
[0029] More refined analysis can be achieved by combining continuous capture data from ground-based infrared cameras. Each valid image captured by the infrared cameras is identified, and the time, location, and identifiable waterbird species information are recorded. By analyzing the number of days waterbirds appear within a specific geographical area (e.g., a grid with 50-meter sides) within a statistical period (e.g., 21 days) and the duration of each appearance, the average daily bird dwell time in that area can be calculated. Areas with an average daily dwell time exceeding a preset threshold (e.g., four hours) are marked by the system as high-frequency waterbird activity areas.
[0030] These areas are often important foraging grounds, resting places, or roosting sites for birds.
[0031] S1-2. Determine the coordinates of specific observation points within the safe observation range. After identifying high-frequency activity areas, observation facilities are not directly built in the center of the area. Instead, the principle of minimizing ecological disturbance must be followed. Based on the ecological habits and sensitivities of the target waterbird species, a safe observation distance is set. Using this distance as the radius, a ring-shaped buffer zone is delineated around the high-frequency activity area. The final observation points will be selected from within this buffer zone.
[0032] When selecting a site, multiple on-site conditions need to be comprehensively evaluated. For example, the geological conditions must be stable, the soil bearing capacity must meet the requirements for long-term installation of the facility, and areas that are easily flooded by tides, such as tidal gullies, must be avoided. The field of view must be open, and digital elevation models and field of view analysis tools must be used to ensure that most of the target high-frequency activity area can be effectively observed from the candidate site, avoiding serious obstruction caused by vegetation or other terrain features.
[0033] In addition, the feasibility of equipment transportation and installation must be considered, as well as keeping it as far away as possible from existing pedestrian walkways or patrol roads to reduce the cumulative disturbance to birds from different sources of human activity.
[0034] After the above systematic evaluation, the operators use measuring equipment to determine the final latitude and longitude coordinates of each observation point, thereby forming a layout plan of the target wetland bird observation points.
[0035] S2. Construct a model observation station at the observation point; Considering that the construction of the observation station needs to take into account the environmental requirements of the observation point as well as the requirements for construction and transportation, the observation station in this step can be a modular green birdwatching hut. The core is to build a modular hardware platform that is highly adaptable to the environment, self-sufficient in energy, has low ecological disturbance, and is easy to deploy.
[0036] In some embodiments, the smart birdwatching hut of the present invention adopts a fully modular design concept, and its structural composition and connection relationship are as follows: Figure 2 As shown.
[0037] The main structure can be mainly composed of a foundation support module 1, a wall enclosure module 2, an observation window module 3, and a roof energy integration module 4.
[0038] The basic support module can use adjustable feet and a stable chassis to adapt to the uneven ground at the edge of the wetland, and enhance the overall wind resistance through triangular stabilizers; The wall cladding modules can be constructed from lightweight composite panels resistant to salt spray corrosion. The panels are coated with a special anti-corrosion coating and filled with insulation material. The panels are secured together with waterproof sealing strips and stainless steel connectors to form a well-sealed system. Furthermore, sound-absorbing devices are installed at ventilation openings and other openings in the walls, and sound-absorbing materials can be applied to the interior walls to minimize the transmission of mechanical noise from operating fans, air conditioners, etc. The lighting system is specially designed to use specific wavelengths of infrared light or extremely low-intensity, specific color-temperature visible light for nighttime observation, strictly avoiding light pollution that could disrupt the birds' diurnal rhythms.
[0039] The observation window module can be made of special glass that is resistant to gusts of wind. The outside of the glass can be fitted with a fine grille or coating that has a bird strike protection function. Furthermore, it can also adopt the principle of one-way vision or a clever angle design, so that the observer can clearly observe the outside world from inside the room, while the birds outside the room have difficulty detecting the activities inside the room, thus avoiding the alarm of birds caused by direct human exposure.
[0040] The roof energy integration module can be designed with a suitable slope to maximize sunlight reception and facilitate drainage, and its surface integrates solar panels.
[0041] To avoid impacting the environment of the observation point, the green and low-carbon characteristics of the above-mentioned smart birdwatching huts are mainly reflected in materials, energy and operation. The structural materials are selected from corrosion-resistant, recyclable or environmentally friendly composite materials and stainless steel, reducing the carbon footprint and environmental impact throughout the entire life cycle.
[0042] The core of the rooftop energy integration module is the solar power unit, which converts solar energy into electricity using high-efficiency solar panels installed on the roof. The generated electricity is regulated by an intelligent energy management module. This module features maximum power point tracking to optimize the power generation efficiency of the solar panels. The electricity is stored in dedicated energy storage devices, such as lithium battery packs. The intelligent energy management module continuously monitors the charge status of the energy storage unit, load power requirements, and weather forecasts, and dynamically manages the operating strategies of various electrical devices within the system. When energy storage is insufficient at night or during prolonged periods of cloudy or rainy weather, the system automatically enters a low-power operation mode, temporarily reducing the sampling frequency of non-core sensors or shutting down high-power interactive display devices to prioritize the continuous operation of core monitoring equipment (such as cameras, edge computing units, and communication modules). This ensures that the system can maintain critical functions for up to several weeks even when completely off-grid, achieving true energy self-sufficiency and low-carbon operation.
[0043] During on-site deployment, all prefabricated modules are transported to the selected locations. Construction workers use simple tools to assemble the modules by quickly disassembling the interfaces and using prefabricated connectors, much like building large blocks. The entire process does not require large-scale civil engineering or heavy machinery, and the main structure can be erected and the equipment initially installed in a short time. This minimizes construction damage to wetland surface vegetation and soil structure, achieving rapid and low-impact construction and deployment of the cabin as a whole.
[0044] S3, multimodal data acquisition and local intelligent fusion recognition.
[0045] This step is the core of the technical solution, which realizes the transformation from raw perception to structured cognition by performing real-time intelligent processing at the data source.
[0046] S3-1, Synchronous acquisition of multi-source heterogeneous data.
[0047] A multimodal sensor network deployed in each smart birdwatching hut constitutes the system's sensing front end. This network is a collaborative sensor suite, primarily comprising the following units: The optical imaging unit consists of a high-definition camera with high optical zoom capability, supports autofocus and exposure, and can provide richly detailed color video and images during the day. Some models also have night vision capabilities, which can work in low light conditions through built-in fill light or high-sensitivity sensors. The infrared thermal imaging unit uses the difference in thermal radiation between the target and the background to form an image. Its operation is completely unaffected by visible light conditions. It can clearly display the outline and location of warm-blooded animals at night, in fog, and in rain, effectively making up for the limitations of optical cameras. The acoustic sensing unit, consisting of a highly sensitive acoustic sensor or microphone array, is responsible for continuously collecting environmental sounds within the monitoring area, including bird calls and flapping wing sounds. The environmental parameter monitoring unit integrates multiple sensors to collect meteorological data such as temperature and humidity, atmospheric pressure, light intensity, wind speed and direction around the observation point in real time, as well as optional air quality parameters such as carbon dioxide concentration and volatile organic compound concentration.
[0048] All these sensors are connected to the edge computing unit inside the cabin via a reliable industrial bus or wireless IoT protocol and are precisely time-synchronized to ensure data consistency across the timeline during subsequent fusion analysis.
[0049] S3-2, Intelligent image and sound recognition based on edge computing. Specifically, this step is performed by an embedded information processing device (edge computing unit) deployed locally in the observation cabin. Its core task is to process and analyze the massive amount of raw data collected by the sensors in real time.
[0050] In the bird image recognition process, image preprocessing is performed first to improve image quality and optimize input for subsequent recognition algorithms. Preprocessing operations include, but are not limited to, noise reduction to eliminate image impurities caused by sensor noise, circuit noise, or environmental particles; image enhancement to improve contrast or sharpness and make target features more obvious; and normalization to adjust the image size and pixel value range to a fixed standard required by the recognition model.
[0051] The preprocessed image is input into a pre-trained convolutional neural network model. Through its multi-layered structure of convolutional layers, pooling layers, etc., it can automatically learn and extract deep features related to waterbird species identification from the image, such as feather texture patterns, beak shape and color, leg length and color, overall body shape and posture, etc.
[0052] The model ultimately outputs the identification results of one or more bird targets in the image, including their species, their location bounding box in the image, and a confidence score indicating the degree of certainty.
[0053] In the bird sound recognition process, the system analyzes continuous audio signals collected by acoustic sensors. First, the sound signals undergo preprocessing, including noise reduction to filter out persistent background noise such as wind and water flow; filtering to preserve the typical frequency range of bird calls; and frame segmentation. Next, feature extraction is performed on the preprocessed audio frames. A common method is to transform them from the time domain to the frequency domain, for example, by using a short-time Fourier transform to obtain a spectrogram, and then extracting feature vectors that characterize the essential properties of the sound, such as Mel-frequency cepstral coefficients. These feature vectors are then fed into a machine learning classifier, such as a support vector machine model.
[0054] During the training phase, the support vector machine model has learned a large number of labeled samples of different waterbird calls, which can establish a mapping relationship between sound features and bird species. In the recognition phase, the model determines which bird's call it belongs to based on the features of the input sound and outputs the classification result and confidence level.
[0055] S3-3, Multimodal decision-level fusion and record generation.
[0056] Multimodal data fusion and structured record generation are the final steps of edge intelligence. They can simultaneously receive real-time results from image recognition channels and sound recognition channels, and align them in time with the current environmental sensor readings to ensure that the comparison is between visual and auditory events within the same time period. Then, spatial correlation is established, for example, the directional information of the sound is corroborated by the location information of the target in the image.
[0057] When image recognition and sound recognition make consistent judgments about the same species and both have high confidence levels, the confidence level of the final result is greatly improved. When the two results are inconsistent, the result with higher confidence level is adopted, or a comprehensive decision is made by combining environmental context (such as a certain bird is more likely to make a certain call in a certain season or at a certain time). When only a single modality has reliable detection, the result of that modality is output, but it will be labeled.
[0058] Each successful fusion recognition event, that is, when the appearance of birds is confirmed once or multiple times, will trigger the system to automatically generate a standardized waterbird observation record.
[0059] This record is a structured data object whose key fields include the precise timestamp of the event, the unique identifier of the observation point that generated the record and its fixed geographic coordinates (latitude and longitude), the species of waterbirds identified, the estimated number, and the overall confidence level of the identification results.
[0060] In addition, the record can also be associated with and stored as a thumbnail index of the original image or audio that triggered the recognition, as well as a snapshot of the environmental parameters at the time the record was generated.
[0061] S4. Provide real-time public ecological observation information services.
[0062] This step transforms professional monitoring results into readily perceptible and interactive science popularization experiences for the public, thereby enhancing social participation in wetland conservation.
[0063] The observation service is provided through multiple online and offline channels. Offline, user interaction terminals can be provided, such as a touch screen with an integrated facial recognition module. When a visitor approaches the terminal, it can be activated, and its main interface displays the live birdwatching footage from the main camera of the cabin in real time. Once the edge computing unit generates a high-confidence waterbird observation record, such as identifying a flock of white spoonbills, the system will immediately push a notification to the interactive terminal. A specific area of the terminal screen will dynamically pop up a science card related to the bird species. The card content can include: the bird species' Chinese name, scientific name, picture, protection level in the reserve, a brief introduction to its ecological habits, and behavioral interpretations of the observed behavior.
[0064] Online, a dedicated mobile application or WeChat mini-program can be developed, which the public can download and install on their mobile phones to remotely access the system and apply to view the hut's real-time birdwatching video stream.
[0065] The online content also features intelligent push notifications. Users can subscribe to bird species or specific observation points that interest them. Furthermore, when the system identifies a rare bird species that the user has subscribed to in any of the connected observation points, the app will send a push notification to the user's phone, prompting "The red-crowned crane you are following has appeared at a certain observation point," attracting the user to open and watch immediately.
[0066] By combining online and offline services, the time and space limitations of birdwatching activities have been broken, enabling the achievements of wetland ecological protection to be communicated to the public in real time, thereby enhancing the social influence and public support of the protection work.
[0067] S5, cloud-based big data deep analysis and decision support This step aggregates and mines massive amounts of observation records to produce macroscopic patterns and in-depth knowledge, serving scientific research and conservation management.
[0068] S5-1, Generation of spatial distribution heat map of waterbirds. The cloud platform periodically aggregates waterbird observation records uploaded by all observation huts. For a specific key species, it extracts the geographical coordinates (i.e., observation point coordinates) of the species' appearance in all records. Since each observation point represents a fixed location, a large number of records are spatially represented as the superposition of the frequency or number of events at these points.
[0069] In order to obtain a continuous and smooth distribution pattern, the platform uses spatial clustering algorithms (such as K-means algorithm) to perform cluster analysis on these point data and identify the spatial aggregation centers of bird activities.
[0070] Furthermore, to visualize the distribution density more intuitively, a kernel density estimation method is employed. This method treats each observation point as a heat source, whose heat value is weighted by the number of birds recorded at that point. Then, the cumulative density of each point across the entire geographic area affected by all heat sources is calculated, thus generating a continuous spatial density surface. This density surface is then color-rendered (typically warm colors represent high density, and cool colors represent low density) and transparently overlaid on a high-resolution wetland geographic information system base map.
[0071] The base map contains rich geographic information layers such as water boundaries, vegetation type distribution, roads, settlements, and core and buffer zone boundaries.
[0072] The final composite image generated by overlaying the images is a waterbird distribution heat map. The reserve's managers can use this map to clearly understand the spatial distribution patterns, core habitat ranges, and activity corridors of different bird species in the reserve in real time, historically, or within statistical periods. This provides an extremely intuitive and scientific graphical basis for delineating key patrol areas, assessing habitat quality, planning ecological restoration projects, and managing the scope of tourist activities.
[0073] S5-2, Behavioral pattern recognition based on time-series data.
[0074] For a specific observation point, the platform can sort its long-term observation records by timestamp to form a time series. By analyzing this time series, the rhythmic patterns of bird activity can be revealed. For example, the average frequency of a bird species appearing at different times of the day can be calculated, thus revealing its diurnal activity pattern. The platform can also analyze the changes in the frequency and number of bird appearances in different months and seasons, depicting phenological patterns such as migration season, wintering period, and breeding season.
[0075] More in-depth behavioral analysis requires processing more complex sequence data. The platform can use temporal models in machine learning, such as long short-term memory networks, to model continuous observation data sequences, thereby remembering long-term dependencies and processing sequence data with sequential correlations, such as bird behavior.
[0076] Through training, long short-term memory networks can learn to identify features of specific behavioral patterns from a series of observed states.
[0077] S5-3, Environment-Behavior Correlation Modeling and Prediction. Bird distribution and behavior in wetland ecosystems are influenced by a variety of environmental factors, such as temperature, precipitation, water level, wind speed, and food resource abundance. The cloud platform aligns and integrates historically accumulated waterbird observation datasets (including time, space, species, quantity, and behavioral tags) with corresponding environmental monitoring data for the same time and region in a spatiotemporal dimension.
[0078] Subsequently, predictive models are constructed using fused data, such as random forest algorithms, to quantify and reveal the relationship between environment and behavior. For example, when the water level in the tidal flat area is maintained between 0.8 meters and 1.2 meters and the daily average temperature is higher than 5 degrees Celsius, the foraging activity intensity of wading birds reaches its peak. Alternatively, a northeasterly wind of level 6 or above lasting for more than three days may cause plovers and sandpipers to migrate to the leeward area to the south.
[0079] Based on this well-constructed and validated correlation prediction model, reserve managers can input weather forecasts, hydrological forecasts, and other data for a future period into the system. The model will automatically run simulations and output prediction reports on future bird distribution hotspots, main activity areas, and potential behavioral trends. This data-driven prediction report can provide unprecedented scientific foresight for the dynamic and refined management of the reserve. For example, it can deploy patrol forces to predicted hotspots in advance, scientifically regulate wetland water levels to create suitable habitats, and make advance arrangements for diversion when it is predicted that tourists may overlap with high-density bird areas. This will elevate conservation management from a passive response to a new level of proactive planning.
[0080] In other embodiments, the green and low-carbon coastal wetland smart birdwatching system can be physically and logically divided into four core functional modules, and the modules exchange and control data through standardized interface protocols.
[0081] The observation point deployment selection module corresponds to the relevant content in S1. In actual deployment, the function of this module can be realized by dedicated planning software or hardware with geographic information system functions. Based on the input pre-monitoring data (such as drone aerial images and infrared camera monitoring data), this module assists planners in determining the specific geographical location of the observation point through built-in analysis logic, and provides a location basis for the subsequent construction of modular cabins.
[0082] The data acquisition module is specifically configured at the observation point selected by the observation point deployment module. Its core consists of a series of sensors deployed on or inside the birdwatching hut. These sensors include at least a high-definition optical camera (with zoom and night vision capabilities) for acquiring images of waterbirds, an infrared thermal imaging camera for detecting heat sources, an acoustic sensor for collecting sound, and an ambient air quality sensor for monitoring parameters such as temperature, humidity, carbon dioxide, and volatile organic compounds. These sensors are connected to a local processing unit via a network to collectively acquire monitoring data on images, sound, and environmental parameters.
[0083] The local intelligent processing module, deployed as an edge computing unit inside the birdwatching hut, communicates with the data acquisition module. Its core task is to run the deep learning algorithms described in the disclosure document (such as convolutional neural network models for image recognition and support vector machine models for sound classification) to process the collected multimodal monitoring data in real time, performing a series of operations such as noise reduction, feature extraction, classification and recognition, and association analysis, and finally obtaining the waterbird species identification results.
[0084] The interaction module includes at least two forms: The first is a screen used to display content, providing visitors with real-time data queries, birdwatching footage, and science information; Second, software application interfaces that support remote access (such as the backend services of mobile apps) are used to provide the public with ecological observation information services such as remote birdwatching live streaming and data sharing.
[0085] This application also provides corresponding devices, computer storage media, and computer program products for implementing the technical solutions provided in this application.
[0086] The device includes a memory and a processor. The memory stores instructions or code, and the processor executes the instructions or code to cause the device to perform the data processing method described in any embodiment of this application.
[0087] The computer storage medium stores code, and when the code is run, the device running the code implements the data processing method described in any embodiment of this application.
[0088] The computer program product contains instructions. When run on a computer, it causes the computer to perform the data processing method described in any embodiment of this application.
[0089] In the embodiments of this application, the terms "first" and "second" (if they exist) are used only as name identifiers and do not represent the order of first and second.
[0090] Although this application has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some of the technical features; and these modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of this application.
Claims
1. A green, low-carbon, intelligent birdwatching method for coastal wetlands, characterized in that, Includes the following steps: Based on the environmental parameter data of the target coastal wetland and waterbird activity information, observation points were selected; At the observation point, monitoring data including at least waterbird images, waterbird sounds, and wetland environmental parameters are collected based on a multimodal sensor network. The monitoring data is subjected to local multimodal data fusion and identification processing to obtain waterbird species identification results. The local multimodal data fusion and identification processing includes fusing and analyzing the waterbird images and the waterbird sounds for identification. Based on the waterbird species identification results, ecological observation information services will be provided to the public.
2. The method according to claim 1, characterized in that, The selection of the observation points includes: Based on pre-monitoring data, areas where waterbirds stay for longer than a preset time are identified as high-frequency activity areas for waterbirds. Determine the geographical location of at least one observation point within the high-frequency activity area of waterbirds and / or within a safe observation range outside the high-frequency activity area of waterbirds.
3. The method according to claim 1, characterized in that, The multimodal data fusion and recognition processing includes: The waterbird images are preprocessed with noise reduction and normalization. The sounds of the waterbirds are subjected to noise reduction, filtering preprocessing, and spectrum analysis to extract sound features; A convolutional neural network model was used to identify the species of waterbirds after denoising and normalization, and the image recognition results were obtained. The waterbird sounds were classified and identified using a support vector machine model to obtain the waterbird sound recognition results. The coastal wetland environmental parameter data are correlated with the waterbird images and sounds to obtain the waterbird species identification results.
4. The method according to claim 1, characterized in that, The method further includes: Each event in which a waterbird species is identified based on the monitoring data is associated with the fixed geographic coordinates of the corresponding observation point to generate a waterbird observation record with spatial location information and timestamp information.
5. The method according to claim 1, characterized in that, The method further includes: Upload waterbird observation records with spatial location information from the observation points to the cloud platform; Based on the spatial location information in the waterbird observation records of each observation point, the cloud platform uses a clustering algorithm to perform spatial distribution analysis in order to identify the gathering areas of waterbird activity.
6. The method according to claim 5, characterized in that, The spatial distribution analysis includes: Based on the spatial location information in the waterbird observation records, a spatial density distribution map of waterbird activity is generated using a kernel density estimation algorithm. The spatial density distribution map is overlaid with the geographic information base map of the coastal wetland to generate a waterbird distribution heat map.
7. The method according to claim 5, characterized in that, The method also includes: On the cloud platform, based on the timestamp information and identification results in the waterbird observation records, a time series analysis model is used to identify the behavioral patterns of waterbirds. By combining historical environmental data from observation points with waterbird observation records, a predictive model is used to construct the correlation between environmental parameters and waterbird behavior.
8. A green, low-carbon, intelligent birdwatching system for coastal wetlands, characterized in that, It includes a module for selecting observation points, a data acquisition module, a local intelligent processing module, and an interaction module; The observation point selection module is used to select observation points based on environmental data and bird activity information of the target coastal wetland. The data acquisition module, configured at the observation point selected by the observation point deployment selection module, includes a multimodal sensor network for acquiring monitoring data including at least waterbird images, waterbird sounds, and wetland environmental parameters at the observation point. A local intelligent processing module is communicatively connected to the data acquisition module and is used to perform fusion analysis and identification on the monitoring data to obtain waterbird species identification results. An interactive module, connected to the local intelligent processing module, is used to generate and provide ecological observation information services to the public based on the waterbird species identification results.
9. An electronic device, characterized in that, The electronic device includes at least one processor and at least one memory, the memory being data-connected to the processor, wherein the memory stores instructions executable by the at least one processor, the instructions being executed by the at least one processor to enable the at least one processor to perform the method of any one of claims 1-7.
10. A computer-storable medium, characterized in that, The storable medium stores computer instructions, which, when executed by a processor, specifically perform the steps of the method as described in any one of claims 1-7.
11. A computer program product comprising computer instructions, characterized in that, When the computer instructions are executed by the processor, they specifically perform the steps in the method as described in any one of claims 1-7.