Multi-source AL biodiversity observation method and acquisition system

Abstract

The invention discloses a multi-source AL biodiversity observation method and acquisition system, and relates to the technical field of biodiversity observation, and the system comprises a data acquisition module, a preprocessing module, a species identification module, a data management module and a data display module; the data acquisition module comprises a fixed camera and an unmanned aerial vehicle camera, the fixed camera acquires images of species in a fixed area, and the unmanned aerial vehicle camera covers a large-range area to acquire the images; by combining the fixed camera and the unmanned aerial vehicle camera, data acquisition is not limited by time and space of manual inspection, safe species monitoring and biodiversity analysis are realized, and by combining modern deep learning and data processing technologies, the system can acquire species images, identify species and analyze density in a large area range, and the system has a wide application prospect. Through quantitative analysis of species diversity, an ecologist can rapidly know the ecosystem condition of a certain region and take protection measures in time.

Description

Technical Field

[0001] The present invention relates to the technical field of biodiversity observation, and in particular to a method and collection system for multi-source AL biodiversity observation. Background Art

[0002] Traditional biodiversity observation methods usually rely on manual observation and recording, mainly through field surveys conducted by ecologists or biologists. These methods include visual observation, capture and specimen collection, network surveys, and sample plot surveys. Observers record the species, abundance, and distribution through on-site observations.

[0003] However, traditional biodiversity observation methods are limited by the observer's experience and on-site environmental factors. They cannot fully capture all species, especially small species that are more cryptic. On the other hand, they are not suitable for staff to conduct in-depth research and investigation in some complex and difficult environments.

[0004] In view of the above technical defects, a solution is now proposed. Summary of the Invention

[0005] In view of the shortcomings of the existing technology, the present invention provides a multi-source AL biodiversity observation method and collection system.

[0006] To achieve the above objectives, the present invention is implemented through the following technical solutions: a multi-source AL biodiversity observation method and collection system, including a data collection module, a preprocessing module, a species identification module, a data management module and a data display module;

[0007] The data acquisition module includes fixed cameras and drone cameras. The fixed cameras collect images of species in a fixed area, while the drone cameras cover a large area and collect images. All collected data are transmitted to the preprocessing module;

[0008] The preprocessing module identifies species and deletes irrelevant images, removes noise and enhances species outlines, and transmits the selected and cropped species images to the recognition module;

[0009] The species recognition module uses VGG16 to extract image features, then identifies species information and assigns unique numbers to transmit to the database. Species analysis is performed through KMeans clustering and species density is calculated using Geopandas. The species diversity characteristic value M is calculated based on species density and weight.

[0010] The data management module stores species image data through AWS, uses Django to provide an API interface, uses HTTPS and AES-256 for transmission encryption, regularly backs up data, and stores it in multiple AWS data centers;

[0011] The data display module obtains species categories, locations, and numbers through the API interface, uses Plotly to generate species distribution maps, and uses Matplotl ib to generate density maps to display the spatial distribution and density of species.

[0012] The data acquisition module includes a fixed camera and a drone camera. The fixed camera is installed in a fixed position and is responsible for continuously collecting species image data in a fixed area and marking the time and current camera position information. The drone camera is used to cover a large area. The drone plans the flight path through GPS positioning technology. The drone collects image data by flying and staying at different positions, and marks the time and current camera position information. Finally, all the collected image data is transmitted to the preprocessing module.

[0013] The preprocessing module first uses background detection and object detection algorithms to identify species information in the image, deletes images that do not contain species or only contain background, uses the image similarity detection method SSIM to compare images and remove images with high similarity, uses the denoising algorithm Gaussian filtering to remove noise in the image, and strengthens the species outline through Canny edge detection, then uses the deep learning model Net-50 to extract the basic features of the species, and then uses the R-CNN model to select the species, crop the species image, and remove the environmental pixels around the species. The cropped species image is square in shape, and the cropped image data is transmitted to the image recognition module.

[0014] The species recognition module uses the deep learning model VGG16 to extract features from the cropped image, identifies different species names from the cropped image, and assigns a unique number to each species. The identified species name, location information, and unique number are transmitted to the database management module through the API interface. The KMeans clustering algorithm is used to perform cluster analysis on the identified species, and species with similar characteristics are classified. The data analysis tool Geopandas is used to count the number of species individuals N in a specific ecological area, and then the species density D is calculated based on the area A of the area. The formula is

[0015] The species identification module evaluates the species diversity characteristic value of the ecological region through the species diversity characteristic value M. The calculation formula of the species diversity characteristic value M is as follows:

[0016]

[0017] Among them, Di represents the density of the i-th species, P represents the weight coefficient of the density of the i-th species, S represents the total number of species, and the species diversity characteristic value M is used to measure the level of species diversity. The closer the species diversity characteristic value M is to 1, the lower the species diversity is, and the closer the species diversity characteristic value M is to 0, the higher the species diversity in the area.

[0018] The data management module stores the collected species image data on the cloud platform through Amazon Web Service, provides an API interface through the Django back-end framework, uses the HTTPS encryption protocol for data transmission, and uses the encryption algorithm AES-256 to encrypt the species data stored on the cloud platform. The administrator assigns specific permissions to different users, uses AWS IAM to control API interface access rights, performs data backup regularly, and stores the backup data in multiple AWS data centers.

[0019] The data display module obtains the identified species category, location information and unique number from the data management module through the API interface, uses Plotly to generate a species distribution map to display the spatial distribution of species in the ecological region, and uses Matplotlib to generate a species density map to display the density of species in different areas.

[0020] The specific method of the multi-source AL biodiversity observation collection system is as follows:

[0021] S1, fixed cameras and drone cameras collect species images, mark time and location, and transmit image data to the preprocessing module;

[0022] S2, the preprocessing module filters out species-free or duplicate images, removes noise and enhances species outlines, extracts features using a deep learning model, selects and crops species images, and transmits them to the species recognition module;

[0023] S3, the species identification module uses VGG16 to identify species and assign numbers, and uses KMeans clustering to classify species, thereby calculating species density and diversity characteristic values;

[0024] S4, the data management module stores data on the AWS cloud platform, uses HTTPS and AES-256 encryption, and backs up data regularly;

[0025] S5. The data display module obtains species information through the API, generates distribution maps and density maps, and displays species distribution and density.

[0026] The present invention provides a multi-source AL biodiversity observation method and collection system. Compared with the existing technology, it has the following advantages:

[0027] By combining fixed cameras with drone cameras, the present invention enables data collection to be unrestricted by the time and space limitations of manual inspections, thus enabling safe species monitoring and biodiversity analysis. Combining modern deep learning and data processing technologies, the system can collect species images over large areas and conduct species identification and density analysis. Through quantitative analysis of species diversity, ecologists can quickly understand the status of a region's ecosystem and take timely protective measures. BRIEF DESCRIPTION OF THE DRAWINGS

[0028] Figure 1 It is a schematic diagram of the principle framework of the present invention. DETAILED DESCRIPTION

[0029] The following will clearly and completely describe the technical solutions in the embodiments of the present invention in conjunction with the accompanying drawings. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without making creative efforts are within the scope of protection of the present invention.

[0030] See also Figure 1 ,This application provides a multi-source AL biodiversity observation method and acquisition system, including a data acquisition module, a preprocessing module, a species identification module, a data management module and a data display module;

[0031] The data acquisition module includes fixed cameras and drone cameras. The fixed cameras collect images of species in a fixed area, while the drone cameras cover a large area and collect images. All collected data are transmitted to the preprocessing module;

[0032] The preprocessing module identifies species and deletes irrelevant images, removes noise and enhances species outlines, and transmits the selected and cropped species images to the recognition module;

[0033] The species recognition module uses VGG16 to extract image features, then identifies species information and assigns unique numbers to transmit to the database. Species analysis is performed through KMeans clustering and species density is calculated using Geopandas. The species diversity characteristic value M is calculated based on species density and weight.

[0034] The data management module stores species image data through AWS, uses Django to provide an API interface, uses HTTPS and AES-256 for transmission encryption, regularly backs up data, and stores it in multiple AWS data centers;

[0035] The data display module obtains species categories, locations, and numbers through the API interface, uses Plotly to generate species distribution maps, and uses Matplotl ib to generate density maps to display the spatial distribution and density of species.

[0036] The data acquisition module includes fixed cameras and drone cameras. The fixed cameras are installed in a fixed position and are responsible for continuously collecting species image data in a fixed area and marking the time and current camera position information. The drone cameras are used to cover a large area. The drone plans the flight path through GPS positioning technology. The drone collects image data by flying and staying at different locations, and marks the time and current camera position information. Finally, all collected image data are transmitted to the preprocessing module.

[0037] Drones use GPS positioning technology to plan flight routes, quickly covering larger areas and collecting image data at different heights and angles. Drones can stop at different locations and flexibly adjust flight altitude and angle as needed to obtain more comprehensive images of species from multiple perspectives. Collection time and camera position are also recorded. This mobile collection method can effectively overcome the limitation of fixed cameras that cannot cover a wide area.

[0038] The preprocessing module first uses background detection and object detection algorithms to identify species information in the image, deletes images that do not contain species or only contain background, uses the image similarity detection method SSIM to compare images and remove images with high similarity, uses the denoising algorithm Gaussian filtering to remove noise in the image, and uses Canny edge detection to enhance the species outline. Then, the deep learning model Net-50 is used to extract the basic characteristics of the species, and then the species is framed by the R-CNN model, the species image is cropped, the environmental pixels around the species are removed, the species image is cropped into a square shape, and the cropped image data is transmitted to the image recognition module.

[0039] The species recognition module uses the deep learning model VGG16 to extract features from the cropped image, identifies different species names from the cropped image, and assigns a unique number to each species. The identified species name, location information, and unique number are transmitted to the database management module through the API interface. The KMeans clustering algorithm is used to perform cluster analysis on the identified species, and species with similar characteristics are classified. The data analysis tool Geopandas is used to count the number of species individuals N in a specific ecological area, and then the species density D is calculated based on the area A of the area. The formula is:

[0040] The species identification module evaluates the species diversity characteristic value of the ecological region through the species diversity characteristic value M. The calculation formula of the species diversity characteristic value M is as follows:

[0041]

[0042] Among them, Di represents the density of the i-th species, P represents the weight coefficient of the density of the i-th species, S represents the total number of species, and the species diversity characteristic value M is used to measure the level of species diversity. The closer the species diversity characteristic value M is to 1, the lower the species diversity is, and the closer the species diversity characteristic value M is to 0, the higher the species diversity in the area.

[0043] The data management module stores the collected species image data on the cloud platform through Amazon Web Service, provides an API interface through the Django back-end framework, uses the HTTPS encryption protocol for data transmission, and uses the AES-256 encryption algorithm to encrypt the species data stored on the cloud platform. The administrator assigns specific permissions to different users, uses AWS IAM to control access rights to the API interface, performs data backup regularly, and stores the backup data in multiple AWS data centers.

[0044] The data backup process is automated. The system automatically performs backups according to a predetermined cycle to ensure that all critical data is fully protected. Administrators can view backup records and perform recovery operations as needed. If the original data is damaged or lost, the backup data can help the system quickly resume normal operation.

[0045] The data display module obtains the identified species category, location information and unique number from the data management module through the API interface, uses Plotly to generate a species distribution map to show the spatial distribution of species in the ecological region, and uses Matplotlib to generate a species density map to show the density of species in different areas.

[0046] A multi-source AL biodiversity observation collection system is described in detail below:

[0047] S1, fixed cameras and drone cameras collect species images, mark time and location, and transmit image data to the preprocessing module;

[0048] S2, the preprocessing module filters out species-free or duplicate images, removes noise and enhances species outlines, extracts features using a deep learning model, selects and crops species images, and transmits them to the species recognition module;

[0049] S3, the species identification module uses VGG16 to identify species and assign numbers, and uses KMeans clustering to classify species, thereby calculating species density and diversity characteristic values;

[0050] S4, the data management module stores data on the AWS cloud platform, uses HTTPS and AES-256 encryption, and backs up data regularly;

[0051] S5. The data display module obtains species information through the API, generates distribution maps and density maps, and displays species distribution and density.

[0052] Specific workflow:

[0053] Species image data is collected by fixed cameras and drone cameras. Fixed cameras collect data in specific areas at fixed positions and mark the time and location. UAV cameras cover a large area and then transmit the data to the preprocessing module. The preprocessing module removes images without species and only background, uses similarity detection to remove duplicate images, uses Gaussian filtering to denoise, and uses Canny edge detection to enhance species outlines. Then, species features are extracted through the Net-50 model. Finally, the species image is selected through R-CNN, cropped and transmitted to the species identification module. The species identification module extracts image features through the VGG16 deep learning model to identify the species name and assign a unique number. The KMeans clustering algorithm is used to cluster the species and calculate the species density and diversity characteristic value M. The data display module obtains species information from the data management module through the API interface, uses Plotly to generate a species distribution map to display the spatial distribution of species, and uses Matplotl ib to generate a species density map to display the density of species in different areas.

[0054] Furthermore, the present invention combines fixed cameras with drone cameras, so that data collection is not restricted by the time and space limitations of manual inspections, thereby realizing safe species monitoring and biodiversity analysis. Combined with modern deep learning and data processing technology, the system can collect species images and identify species and analyze density over a large area. Through quantitative analysis of species diversity, ecologists can quickly understand the status of the ecosystem in a certain area and take timely protective measures.

[0055] Some of the data in the above formulas are dimensionless and numerically calculated. Meanwhile, the contents not described in detail in this specification belong to the prior art known to those skilled in the art.

[0056] The above embodiments are only used to illustrate the technical method of the present invention and are not intended to limit the present invention. Although the present invention has been described in detail with reference to the preferred embodiments, those skilled in the art should understand that the technical method of the present invention may be modified or replaced by equivalents without departing from the spirit and scope of the technical method of the present invention.

Claims

1. A multi-source AL biodiversity observation collection system, characterized by: include: Data acquisition module, preprocessing module, species identification module, data management module and data display module; The data acquisition module includes fixed cameras and drone cameras. The fixed cameras collect images of species in a fixed area, while the drone cameras cover a large area and collect images. All collected data are transmitted to the preprocessing module; The preprocessing module identifies species and deletes irrelevant images, removes noise and enhances species outlines, and transmits the selected and cropped species images to the recognition module; The species recognition module uses VGG16 to extract image features, then identifies species information and assigns unique numbers to transmit to the database. Species analysis is performed through KMeans clustering and species density is calculated using Geopandas. The species diversity characteristic value M is calculated based on species density and weight. The data management module stores species image data through AWS, uses Django to provide an API interface, uses HTTPS and AES-256 for transmission encryption, regularly backs up data, and stores it in multiple AWS data centers; The data display module obtains species categories, locations, and numbers through the API interface, uses Plotly to generate species distribution maps, and uses Matplotl ib to generate density maps to display the spatial distribution and density of species.

2. A multi-source AL biodiversity observation and collection system according to claim 1, characterized in that: The data acquisition module includes a fixed camera and a drone camera. The fixed camera is installed in a fixed position and is responsible for continuously collecting species image data in a fixed area and marking the time and current camera position information. The drone camera is used to cover a large area. The drone plans the flight path through GPS positioning technology. The drone collects image data by flying and staying at different positions, and marks the time and current camera position information. Finally, all the collected image data is transmitted to the preprocessing module.

3. The multi-source AL biodiversity observation and collection system according to claim 1, characterized in that: The preprocessing module first uses background detection and object detection algorithms to identify species information in the image, deletes images that do not contain species or only contain background, uses the image similarity detection method SSIM to compare images and remove images with high similarity, uses the denoising algorithm Gaussian filtering to remove noise in the image, and strengthens the species outline through Canny edge detection, then uses the deep learning model Net-50 to extract the basic features of the species, and then uses the R-CNN model to select the species, crop the species image, and remove the environmental pixels around the species. The cropped species image is square in shape, and the cropped image data is transmitted to the image recognition module.

4. The multi-source AL biodiversity observation and collection system according to claim 1, characterized in that: The species recognition module uses the deep learning model VGG16 to extract features from the cropped image, identifies different species names from the cropped image, and assigns a unique number to each species. The identified species name, location information, and unique number are transmitted to the database management module through the API interface. The KMeans clustering algorithm is used to perform cluster analysis on the identified species, and species with similar characteristics are classified. The data analysis tool Geopandas is used to count the number of species individuals N in a specific ecological area, and then the species density D is calculated based on the area A of the area. The formula is 5. The multi-source AL biodiversity observation and collection system according to claim 1, characterized in that: The species identification module evaluates the species diversity characteristic value of the ecological region through the species diversity characteristic value M. The calculation formula of the species diversity characteristic value M is as follows: Among them, D i represents the density of the i-th species, P represents the weight coefficient of the density of the i-th species, S represents the total number of species, and the species diversity characteristic value M is used to measure the level of species diversity. The closer the species diversity characteristic value M is to 1, the lower the species diversity is. The closer the species diversity characteristic value M is to 0, the higher the species diversity in the area is.

6. The multi-source AL biodiversity observation and collection system according to claim 1, characterized in that: The data management module stores the collected species image data on the cloud platform through Amazon Web Service, provides an API interface through the Django back-end framework, uses the HTTPS encryption protocol for data transmission, and uses the encryption algorithm AES-256 to encrypt the species data stored on the cloud platform. The administrator assigns specific permissions to different users, uses AWS IAM to control API interface access rights, performs data backup regularly, and stores the backup data in multiple AWS data centers.

7. The multi-source AL biodiversity observation and collection system according to claim 1, characterized in that: The data display module obtains the identified species category, location information and unique number from the data management module through the API interface, uses Plotly to generate a species distribution map to display the spatial distribution of species in the ecological region, and uses Matplotl ib to generate a species density map to display the density of species in different areas.

8. A multi-source AL biodiversity observation and collection system, based on the multi-source AL biodiversity observation and collection system according to any one of claims 1 to 7, characterized in that: The specific method of the multi-source AL biodiversity observation collection system is as follows: S1, fixed cameras and drone cameras collect species images, mark time and location, and transmit image data to the preprocessing module; S2, the preprocessing module filters out species-free or duplicate images, removes noise and enhances species outlines, extracts features using a deep learning model, selects and crops species images, and transmits them to the species recognition module; S3, the species identification module uses VGG16 to identify species and assign numbers, and uses KMeans clustering to classify species, thereby calculating species density and diversity characteristic values; S4, the data management module stores data on the AWS cloud platform, uses HTTPS and AES-256 encryption, and backs up data regularly; S5. The data display module obtains species information through the API, generates distribution maps and density maps, and displays species distribution and density.

Images