CN116740646A

CN116740646A - Group identification method and system for monitoring waiting bird habitat

Info

Publication number: CN116740646A
Application number: CN202310830942.1A
Authority: CN
Inventors: 郑福超; 方朝阳; 矢佳昱; 徐骏峰; 肖昕
Original assignee: Jiangxi Normal University
Current assignee: Jiangxi Normal University
Priority date: 2023-07-07
Filing date: 2023-07-07
Publication date: 2023-09-12

Abstract

The invention provides a group identification method and a system for monitoring a waiting habitat, wherein the method comprises the following steps: dividing the bird waiting monitoring image into small images with preset parts; inputting the small images into a trained waiting bird target recognition model, and respectively marking each target area in the small images for the first time according to a target recognition result; judging whether a repeated target area exists in the overlapping area according to the contour coordinates of the overlapping area; if the overlapping area has the repeated target area, deleting one of the repeated target areas contained in any two adjacent small images; inputting an image to be identified into a trained fine-grained identification model of the waiting birds, and obtaining the types respectively corresponding to each waiting bird target respectively included in each target area; and carrying out secondary marking on each waiting bird target respectively included in each target area, and summarizing secondary marking results in all images to be identified. The method and the device can accurately identify the types of the targets of each waiting bird contained in each target area, and further accurately calculate the total number of the types of each waiting bird in the target habitat.

Description

Group identification method and system for monitoring waiting bird habitat

Technical Field

The invention relates to the technical field of biological recognition, in particular to a group recognition method and system for monitoring a waiting bird habitat.

Background

At present, with the progress and development of human society, the awareness of human for natural ecological environment protection is continuously enhanced, birds have been listed as protection objects as friends of human beings, and thus great significance is brought to the research of birds.

The migration of the bird waiting is an adaptive behavior for periodically changing environmental factors, has important significance for population reproduction, evolution and biodiversity maintenance, and the traditional method mainly has two kinds, namely, the traditional method is realized by manual observation, but the identification and monitoring modes of the rare bird waiting are realized by manual observation, so that the problems of low efficiency, high cost and the like exist. The other is that the kind and the number of the birds to be identified can be identified by combining a deep learning technology, but a large number of bird samples are required to be supported by the identification precision, a large amount of manpower is required to be consumed by utilizing the platform video monitoring to intercept the bird sample which is the most accurate in real time, and the identification precision is still not up to an ideal effect due to the influence of the migration time and the bird activity of the birds to be identified.

Disclosure of Invention

Based on the above, the invention aims to provide a group identification method and a system for monitoring a waiting habitat, which aim to solve the problem of lower accuracy in the traditional method for monitoring and identifying the waiting habitat.

The group identification method for monitoring the habitat of the waiting birds, which is provided by the invention, is applied to a waiting bird monitoring and management platform, and comprises the following steps:

acquiring a bird waiting monitoring image of a target habitat at intervals of a first preset time, and dividing the bird waiting monitoring image into small images with preset parts, wherein any two adjacent small images have overlapping parts;

inputting the small images into a trained waiting target recognition model, so as to respectively mark each target area in the small images for the first time according to a target recognition result, wherein each target area at least comprises one waiting target;

acquiring contour coordinates of an overlapping area between any two adjacent small images, and judging whether a repeated target area exists in the overlapping area according to the contour coordinates of the overlapping area;

if the overlapped area has the repeated target area, deleting one of the repeated target areas contained in any two adjacent small images to obtain an image to be identified after deleting the repeated target area;

inputting the images to be identified corresponding to each small image into a trained fine-granularity recognition model of the bird waiting, so that the fine-granularity recognition model of the bird waiting recognizes each first marked target area in the images to be identified, and each target area comprises a corresponding category of each bird waiting target;

and carrying out secondary marking on each bird waiting target respectively included in each target area according to the types respectively corresponding to each bird waiting target respectively included in each target area, and summarizing secondary marking results in all the images to be identified to obtain total numbers respectively corresponding to each bird waiting type.

In summary, according to the group identification method for monitoring the bird waiting habitat, target area locking is firstly performed on the bird waiting monitoring image with the ultra-large memory, and then targeted target type identification is performed on each target area, so that the types of each bird waiting target contained in each target area are accurately identified, and the total number of each bird waiting type in the target habitat is accurately calculated. Specifically, firstly, real-time monitoring a bird waiting monitoring image of a target habitat, dividing the bird waiting monitoring image into a plurality of small images, simultaneously ensuring that the adjacent small images are overlapped, then carrying out target area identification on the small images, marking the identified target area for the first time, judging whether any overlapping area has a repeated target area, if so, only reserving one target area, inputting the image to be identified, from which the repeated target area is deleted, into a bird waiting fine-grained identification model, and further identifying the type of each bird waiting target contained in each target area, thereby accurately obtaining the total number of each bird waiting type.

Further, the step of acquiring the bird waiting monitoring image of the target habitat at intervals of a first preset time, and dividing the bird waiting monitoring image into small images with preset parts, wherein any two adjacent small images have overlapping parts comprises the following steps:

acquiring size information of the bird waiting monitoring image, acquiring position coordinates of all first cutting lines according to the size information and a first preset interval value, acquiring position coordinates of all second cutting lines according to the position coordinates of the first cutting lines and a second preset value, and forming a plurality of overlapping areas by the first cutting lines and the second cutting lines;

and dividing the bird waiting monitoring image into a plurality of small images according to the position coordinates of all the first cutting lines and the position coordinates of all the second cutting lines.

Further, the step of inputting the small image into a trained candidate target recognition model to respectively mark each target area in the small image for the first time according to a target recognition result, wherein each target area at least comprises one candidate target comprises the following steps:

acquiring a plurality of historical bird waiting images, and acquiring position information of all known target areas contained in each historical bird waiting image according to the historical bird waiting images;

labeling each historical bird waiting image according to the position information of all known target areas contained in each historical bird waiting image, and training the bird waiting target recognition model according to the labeled historical bird waiting images.

Further, the step of acquiring the contour coordinates of the overlapping area between any two adjacent small images and judging whether a repeated target area exists in the overlapping area according to the contour coordinates of the overlapping area includes:

and acquiring the contour coordinates of each target area marked for the first time, traversing all the target areas according to the contour coordinates of the overlapped areas and the contour coordinates of each target area, and judging whether the target areas exist in the overlapped areas or not.

Further, the step of inputting the images to be identified corresponding to each small image into a trained fine-granularity recognition model of a bird waiting for the bird to be identified, so that the fine-granularity recognition model of the bird waiting for the bird identifies each first marked target area in the images to be identified, and the step of obtaining the category corresponding to each bird waiting target respectively included in each target area includes:

obtaining a bird waiting image of a known bird waiting type, wherein each bird waiting target in the bird waiting image corresponds to one bird waiting type, and extracting characteristics of each bird waiting target in the bird waiting image to obtain head characteristic information, neck characteristic information, trunk characteristic information and leg characteristic information corresponding to each bird waiting target;

training the fine-grained recognition model of the candidate bird according to head characteristic information, neck characteristic information, trunk characteristic information and leg characteristic information corresponding to each candidate bird target and the candidate bird type corresponding to each candidate bird target;

and identifying each target area in the image to be identified according to the head characteristic information, the neck characteristic information, the trunk characteristic information and the leg characteristic information by the trained fine-granularity waiting bird identification model.

In another aspect the present invention provides a group identification system for monitoring a waiting habitat, the system comprising:

the image segmentation module is used for acquiring a bird waiting monitoring image of a target habitat at intervals of a first preset time, and segmenting the bird waiting monitoring image into small images with preset parts, wherein any two adjacent small images have overlapping parts;

the first marking module is used for inputting the small images into a trained waiting target recognition model so as to respectively mark each target area in the small images for the first time according to a target recognition result, wherein each target area at least comprises one waiting target;

the repeated area detection module is used for acquiring the contour coordinates of the overlapped area between any two adjacent small images and judging whether a repeated target area exists in the overlapped area according to the contour coordinates of the overlapped area;

the repeated region deleting module is used for deleting one repeated target region contained in any two adjacent small images if the repeated target region exists in the overlapped region, so as to obtain an image to be identified after deleting the repeated target region;

the waiting type recognition module is used for inputting the images to be recognized corresponding to each small image into a trained waiting fine granularity recognition model so that the waiting fine granularity recognition model recognizes each first marked target area in the images to be recognized to obtain a type corresponding to each waiting target respectively included in each target area;

and the secondary marking module is used for carrying out secondary marking on each bird waiting target respectively included in each target area according to the type respectively corresponding to each bird waiting target respectively included in each target area, and summarizing secondary marking results in all the images to be identified to obtain the total number respectively corresponding to each bird waiting type.

Further, the image segmentation module further includes:

the cutting line acquisition unit is used for acquiring the size information of the bird waiting monitoring image, acquiring the position coordinates of all first cutting lines according to the size information and a first preset interval value, acquiring the position coordinates of all second cutting lines according to the position coordinates of the first cutting lines and a second preset value, and forming a plurality of overlapping areas by the first cutting lines and the second cutting lines;

and the cutting execution unit is used for dividing the bird waiting monitoring image into a plurality of small images according to the position coordinates of all the first cutting lines and the position coordinates of all the second cutting lines.

Further, the first marking module further includes:

the historical bird waiting image acquisition unit is used for acquiring a plurality of historical bird waiting images and acquiring the position information of all known target areas contained in each historical bird waiting image according to the historical bird waiting images;

the historical bird waiting image labeling unit is used for labeling each historical bird waiting image according to the position information of all known target areas contained in each historical bird waiting image, and training the bird waiting target recognition model according to the labeled historical bird waiting images.

Further, the repeated area detection module further includes:

a contour coordinate obtaining unit, configured to obtain contour coordinates of each target area marked for the first time, and traverse all the target areas according to the contour coordinates of the overlapping areas and the contour coordinates of each target area, so as to determine whether there is a target area in the overlapping areas

Further, the waiting bird species recognition module further includes:

the characteristic extraction unit is used for obtaining a bird waiting image of a known bird waiting type, each bird waiting target in the bird waiting image corresponds to one bird waiting type, and characteristic extraction is carried out on each bird waiting target in the bird waiting image to obtain head characteristic information, neck characteristic information, trunk characteristic information and leg characteristic information corresponding to each bird waiting target;

the fine-granularity recognition model training unit is used for training the fine-granularity recognition model according to the head characteristic information, the neck characteristic information, the trunk characteristic information and the leg characteristic information corresponding to each candidate bird target and the candidate bird type corresponding to each candidate bird target;

and the waiting bird type recognition execution unit is used for recognizing each target area in the image to be recognized according to the head characteristic information, the neck characteristic information, the trunk characteristic information and the leg characteristic information by the trained waiting bird fine granularity recognition model.

In another aspect, the present invention provides a storage medium, including the storage medium storing one or more programs that when executed implement a group identification method for monitoring a waiting habitat as described above.

Another aspect of the invention also provides a computer device comprising a memory and a processor, wherein:

the memory is used for storing a computer program;

the processor is configured to implement the group identification method for monitoring the habitat of a waiting bird as described above when executing the computer program stored on the memory.

Additional aspects and advantages of the invention will be set forth in part in the description which follows and, in part, will be obvious from the description, or may be learned by practice of the invention.

Drawings

FIG. 1 is a flow chart of a method for identifying a community for monitoring a habitat of a waiting bird according to a first embodiment of the present invention;

FIG. 2 is a schematic diagram of identifying a bird candidate according to a first embodiment of the present invention;

FIG. 3 is a diagram showing the result of identifying the species of a bird in the first embodiment of the present invention;

fig. 4 is a schematic structural diagram of a group identification system for monitoring a habitat of a waiting bird according to a second embodiment of the present invention.

The invention will be further described in the following detailed description in conjunction with the above-described figures.

Detailed Description

In order that the invention may be readily understood, a more complete description of the invention will be rendered by reference to the appended drawings. Several embodiments of the invention are shown in the drawings. This invention may, however, be embodied in many different forms and should not be construed as limited to the embodiments set forth herein. Rather, these embodiments are provided so that this disclosure will be thorough and complete.

Unless defined otherwise, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this invention belongs. The terminology used herein in the description of the invention is for the purpose of describing particular embodiments only and is not intended to be limiting of the invention. The term "and/or" as used herein includes any and all combinations of one or more of the associated listed items.

Referring to fig. 1, a flowchart of a method for monitoring a bird habitat according to a first embodiment of the present invention is shown, the method includes steps S01 to S06, wherein:

step S01: acquiring a bird waiting monitoring image of a target habitat at intervals of a first preset time, and dividing the bird waiting monitoring image into small images with preset parts, wherein any two adjacent small images have overlapping parts;

it should be noted that, the target habitat is generally a place where birds are easy to perch in a lakeside or a wetland, the bird waiting monitoring image is obtained by shooting by a camera installed near the habitat, meanwhile, because the target habitat generally occupies a larger space, the bird waiting monitoring image obtained by shooting is generally an image with an ultra-large memory, in order to improve the recognition efficiency and the recognition accuracy, the bird waiting monitoring image is firstly divided into a plurality of small images, and meanwhile, in order to avoid data loss and repeated counting, the overlapping part of each small image needs to be ensured in the dividing process.

In some embodiments of the present invention, in order to divide a bird waiting monitoring image, first, size information of the bird waiting monitoring image is obtained, position coordinates of all first cutting lines are obtained according to the size information and a first preset interval value, position coordinates of all second cutting lines are obtained according to the position coordinates of the first cutting lines and a second preset value, the first cutting lines and the second cutting lines form a plurality of overlapping areas, and it is noted that a plurality of first cutting lines and second cutting lines exist in a transverse direction and a longitudinal direction, and an area between adjacent first cutting lines and second cutting lines is an overlapping portion.

And then dividing the bird waiting monitoring image into a plurality of small images according to the position coordinates of all the first cutting lines and the position coordinates of all the second cutting lines. If there is no overlapping part during the division, the accuracy of identifying the species is easily affected by the data loss, and if the same waiting target is divided into two parts, the repeated waiting target is not easily eliminated.

The first preset time is set to monitor the target habitat once a day or a week, that is, a day or a week, in general, in order to continuously monitor the target habitat in real time.

Step S02: inputting the small images into a trained waiting target recognition model, so as to respectively mark each target area in the small images for the first time according to a target recognition result, wherein each target area at least comprises one waiting target;

in this step, in order to construct a trained target recognition model of a candidate bird, a plurality of historical candidate bird images are first acquired, position information of all known target areas included in each historical candidate bird image is acquired according to the historical candidate bird images, then each historical candidate bird image is labeled according to the position information of all known target areas included in each historical candidate bird image, and the target recognition model of a candidate bird is trained according to the labeled historical candidate bird images.

Referring to fig. 2, a schematic diagram of identifying a candidate bird target is shown, and as can be seen from fig. 2, all areas with a candidate bird target in an image identified according to a trained candidate bird target identification model are identified, and then the identified areas are marked for the first time by using a square frame.

Furthermore, it should be further noted that, because in the actual recognition process, a plurality of next waiting birds may be identified as a waiting bird target due to the problem of shooting angle, which affects accuracy of category counting.

Step S03: acquiring contour coordinates of an overlapping area between any two adjacent small images, and judging whether a repeated target area exists in the overlapping area according to the contour coordinates of the overlapping area;

it should be noted that, in order to avoid the repetition of the count caused by the overlapping portion, it is necessary to accurately obtain the contour coordinate of each target area of the first mark, that is, the contour coordinate of each mark frame of the first mark, and traverse all the target areas according to the contour coordinate of the overlapping area and the contour coordinate of each target area, so as to determine whether there is a target area in the overlapping area, where the contour coordinate of the overlapping area is obtained from the position coordinates of the first cutting line and the second cutting line that constitute the overlapping area.

Step S04: if the overlapped area has the repeated target area, deleting one of the repeated target areas contained in any two adjacent small images to obtain an image to be identified after deleting the repeated target area;

it will be appreciated that when overlapping regions are identified, only one of the target regions is correspondingly reserved, so as to avoid repeated identification in the subsequent category identification process.

Step S05: inputting the images to be identified corresponding to each small image into a trained fine-granularity recognition model of the bird waiting, so that the fine-granularity recognition model of the bird waiting recognizes each first marked target area in the images to be identified, and each target area comprises a corresponding category of each bird waiting target;

in order to accurately identify the type of each candidate bird object in the image to be identified, obtaining a candidate bird image of a known candidate bird type, wherein each candidate bird object in the candidate bird image corresponds to one candidate bird type, and extracting features of each candidate bird object in the candidate bird image to obtain head feature information, neck feature information, trunk feature information and leg feature information corresponding to each candidate bird object;

training the fine-grained recognition model of the candidate bird according to head characteristic information, neck characteristic information, trunk characteristic information and leg characteristic information corresponding to each candidate bird target and the candidate bird type corresponding to each candidate bird target; the head characteristic information, the neck characteristic information, the trunk characteristic information and the leg characteristic information of the waiting bird are integrated into the species identification, so that the species identification accuracy of the waiting bird can be greatly improved.

And finally, identifying each target area in the image to be identified by utilizing the trained fine-grained identification model of the waiting bird according to the head characteristic information, the neck characteristic information, the trunk characteristic information and the leg characteristic information.

Step S06: and carrying out secondary marking on each bird waiting target respectively included in each target area according to the types respectively corresponding to each bird waiting target respectively included in each target area, and summarizing secondary marking results in all the images to be identified to obtain total numbers respectively corresponding to each bird waiting type.

It should be noted that, in this step, referring to fig. 3, a schematic diagram of a candidate bird type recognition result is shown, and since the secondary marked object is each candidate bird target, the marking result is the candidate bird type of each identified candidate bird target, so only all secondary marking frames and candidate bird types corresponding to the secondary marking frames need to be summarized, and the total number of each candidate bird type can be obtained.

Referring to fig. 4, a schematic diagram of a group identification system for monitoring a habitat of a waiting bird according to a second embodiment of the present invention is shown, the system comprising:

the image segmentation module 10 is used for acquiring a bird waiting monitoring image of a target habitat at intervals of a first preset time, and segmenting the bird waiting monitoring image into small images with preset parts, wherein any two adjacent small images have overlapping parts;

further, the image segmentation module 10 further includes:

The first marking module 20 is configured to input the small image into a trained candidate target recognition model, so as to respectively mark each target area in the small image for the first time according to a target recognition result, where each target area includes at least one candidate target;

further, the first marking module 20 further includes:

The repeated area detection module 30 is configured to obtain the contour coordinates of the overlapping area between any two adjacent small images, and determine whether a repeated target area exists in the overlapping area according to the contour coordinates of the overlapping area;

further, the repeated area detection module 30 further includes:

and the contour coordinate acquisition unit is used for acquiring the contour coordinate of each target area marked for the first time, traversing all the target areas according to the contour coordinate of the overlapped area and the contour coordinate of each target area, and judging whether the target area exists in the overlapped area.

The repeated region deleting module 40 is configured to delete, if there is a repeated target region in the overlapping region, one of the target regions with the repeated target regions included in any two adjacent small images, so as to obtain an image to be identified after deleting the target region with the repeated target region;

the waiting type recognition module 50 is configured to input the images to be recognized corresponding to each small image into a trained waiting fine granularity recognition model, so that the waiting fine granularity recognition model recognizes each first marked target area in the images to be recognized, and obtains a type corresponding to each waiting target included in each target area;

further, the waiting bird species recognition module 50 further includes:

The secondary marking module 60 is configured to perform secondary marking on each candidate bird target included in each target area according to the category corresponding to each candidate bird target included in each target area, and aggregate the secondary marking results in all the images to be identified, so as to obtain the total number corresponding to each candidate bird category.

In another aspect, the present invention also provides a storage medium having one or more programs stored thereon, which when executed by a processor, implement the above-described method of group identification for monitoring a habitat of a waiting bird.

In another aspect, the present invention also provides a computer device, including a memory and a processor, where the memory is configured to store a computer program, and the processor is configured to execute the computer program stored on the memory, so as to implement the above group identification method for monitoring a habitat of a waiting bird.

Those of skill in the art will appreciate that the logic and/or steps represented in the flow diagrams or otherwise described herein, e.g., a ordered listing of executable instructions for implementing logical functions, can be embodied in any computer-readable medium for use by or in connection with an instruction execution system, apparatus, or device, such as a computer-based system, processor-containing system, or other system that can fetch the instructions from the instruction execution system, apparatus, or device and execute the instructions. For the purposes of this description, a "computer-readable medium" can be any means that can contain or store the program for use by or in connection with the instruction execution system, apparatus, or device.

More specific examples (a non-exhaustive list) of the computer-readable medium would include the following: an electrical connection (electronic device) having one or more wires, a portable computer diskette (magnetic device), a Random Access Memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM or flash memory), an optical fiber 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, as the program may be electronically captured, via, for instance, optical scanning of the paper or other medium, then compiled, interpreted or otherwise processed in a suitable manner, if necessary, and then stored in a computer memory.

It is to be understood that portions of the present invention may be implemented in hardware, software, firmware, or a combination thereof. In the above-described embodiments, the various steps or methods may be implemented in software or firmware stored in a memory and executed by a suitable instruction execution system. For example, if implemented in hardware, as in another embodiment, may be implemented using any one or combination of the following techniques, as is well known in the art: discrete logic circuits having logic gates for implementing logic functions on data signals, application specific integrated circuits having suitable combinational logic gates, programmable Gate Arrays (PGAs), field Programmable Gate Arrays (FPGAs), and the like.

In the description of the present specification, a description referring to terms "one embodiment," "some embodiments," "examples," "specific examples," or "some examples," etc., means that a particular feature, structure, material, or characteristic described in connection with the 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 embodiments or examples. Furthermore, the particular features, structures, materials, or characteristics described may be combined in any suitable manner in any one or more embodiments or examples.

The above examples merely represent a few embodiments of the present invention, which are described in more detail and are not to be construed as limiting the scope of the present invention. It should be noted that it will be apparent to those skilled in the art that several variations and modifications can be made without departing from the spirit of the invention, which are all within the scope of the invention. Accordingly, the scope of the invention should be assessed as that of the appended claims.

Claims

1. A group identification method for monitoring a waiting habitat, which is applied to a waiting monitoring management platform, and is characterized by comprising the following steps:

2. The method for identifying a population for monitoring a waiting habitat of claim 1, wherein the steps of acquiring a waiting monitor image of a target habitat every first preset time and dividing the waiting monitor image into a preset number of small images, each of any two adjacent small images having an overlapping portion, comprise:

3. The method of claim 1, wherein the step of inputting the small images into a trained candidate target recognition model to respectively first mark each target area in the small images according to target recognition results, each target area including at least one candidate target comprises:

4. The method for monitoring a community of waiting habitats as claimed in claim 2, wherein the step of acquiring the contour coordinates of the overlapping area between any two adjacent small images and judging whether there is a repeated target area in the overlapping area according to the contour coordinates of the overlapping area comprises:

5. The method for identifying a population for monitoring a waiting habitat according to claim 1, wherein the step of inputting the images to be identified corresponding to each of the small images into a trained fine-grained waiting identification model to enable the fine-grained waiting identification model to identify each first-marked target area in the images to be identified, and obtaining a category corresponding to each of the targets of the waiting respectively included in each of the target areas includes:

6. A group identification system for monitoring a waiting habitat, the system comprising:

7. The population identification system for monitoring a waiting habitat of claim 6 wherein said image segmentation module further comprises:

8. The group identification system for monitoring a waiting habitat of claim 6 wherein said first time marking module further comprises:

9. The community identification system of monitoring a waiting habitat of claim 7, wherein said repeat area detection module further comprises:

10. The group identification system for monitoring a waiting habitat of claim 6, wherein said waiting species identification module further comprises: