CN111709421B - Bird identification method, bird identification device, computer equipment and storage medium - Google Patents
Bird identification method, bird identification device, computer equipment and storage medium Download PDFInfo
- Publication number
- CN111709421B CN111709421B CN202010557975.XA CN202010557975A CN111709421B CN 111709421 B CN111709421 B CN 111709421B CN 202010557975 A CN202010557975 A CN 202010557975A CN 111709421 B CN111709421 B CN 111709421B
- Authority
- CN
- China
- Prior art keywords
- bird
- training
- model
- image
- detection
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Active
Links
Images
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V10/00—Arrangements for image or video recognition or understanding
- G06V10/20—Image preprocessing
- G06V10/26—Segmentation of patterns in the image field; Cutting or merging of image elements to establish the pattern region, e.g. clustering-based techniques; Detection of occlusion
- G06V10/267—Segmentation of patterns in the image field; Cutting or merging of image elements to establish the pattern region, e.g. clustering-based techniques; Detection of occlusion by performing operations on regions, e.g. growing, shrinking or watersheds
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/21—Design or setup of recognition systems or techniques; Extraction of features in feature space; Blind source separation
- G06F18/214—Generating training patterns; Bootstrap methods, e.g. bagging or boosting
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/24—Classification techniques
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/04—Architecture, e.g. interconnection topology
- G06N3/045—Combinations of networks
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/08—Learning methods
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V20/00—Scenes; Scene-specific elements
- G06V20/50—Context or environment of the image
- G06V20/52—Surveillance or monitoring of activities, e.g. for recognising suspicious objects
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V2201/00—Indexing scheme relating to image or video recognition or understanding
- G06V2201/07—Target detection
Landscapes
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Physics & Mathematics (AREA)
- Data Mining & Analysis (AREA)
- General Physics & Mathematics (AREA)
- Life Sciences & Earth Sciences (AREA)
- Artificial Intelligence (AREA)
- General Engineering & Computer Science (AREA)
- Evolutionary Computation (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Computational Linguistics (AREA)
- Bioinformatics & Computational Biology (AREA)
- Bioinformatics & Cheminformatics (AREA)
- Health & Medical Sciences (AREA)
- Biomedical Technology (AREA)
- Biophysics (AREA)
- Evolutionary Biology (AREA)
- General Health & Medical Sciences (AREA)
- Molecular Biology (AREA)
- Computing Systems (AREA)
- Mathematical Physics (AREA)
- Software Systems (AREA)
- Multimedia (AREA)
- Image Analysis (AREA)
Abstract
The invention relates to a bird recognition method, a device, a computer device and a storage medium, wherein the method comprises the steps of obtaining an image shot by a panoramic network camera to obtain a panoramic image; inputting the panoramic image into a bird detection model for bird detection to obtain a detection result; judging whether the detection result is a result of birds; if the detection result is that birds exist, acquiring bird detail images from the ball machine, and dividing the bird detail images by adopting the detection result to obtain detail images; inputting the detail image into a bird recognition model for bird recognition to obtain a recognition result; and storing the identification result and the detail image into a database. The bird identification device realizes automatic bird identification, improves identification efficiency and accuracy, does not have monitoring blind areas, and has low cost.
Description
Technical Field
The present invention relates to a bird recognition method, and more particularly, to a bird recognition method, apparatus, computer device, and storage medium.
Background
Birds are important indicators for biodiversity monitoring and ecological environmental impact evaluation. The current situation of bird resources can be known through investigation and monitoring of bird species, the characteristics of the composition, the number, the diversity and the like of the bird species can be summarized, and the characteristics can be utilized to directly reflect the environmental quality of habitats, the health degree of an ecological system, the biodiversity condition, the interference degree of human activities on the ecological system, the influence degree of land utilization and landscape change on the ecological system and the like, so that the birds are required to be identified and supervised, and the birds and the human can be ensured to be in harmony.
The existing bird identification method adopts a single-camera or ball-robbing linkage mode to shoot corresponding photos and then adopts a manual classification identification method to carry out bird identification, but the mode needs to consume a large amount of manpower, material resources and financial resources, in addition, a monitoring blind area exists, all the photos of birds in the whole monitoring area cannot be shot, if the whole monitoring area needs to be covered, a large number of monitoring cameras with fixed visual angles need to be arranged, and the cost is high.
Therefore, a new method is necessary to be designed, so that birds can be automatically identified, the identification efficiency and accuracy are improved, no monitoring blind area exists, and the cost is low.
Disclosure of Invention
The invention aims to overcome the defects of the prior art and provides a bird identification method, a bird identification device, computer equipment and a storage medium.
In order to achieve the above purpose, the present invention adopts the following technical scheme: a method of bird identification comprising:
acquiring an image shot by a panoramic network camera to obtain a panoramic image;
inputting the panoramic image into a bird detection model for bird detection to obtain a detection result;
judging whether the detection result is a result of birds;
if the detection result is that birds exist, acquiring bird detail images from the ball machine, and dividing the bird detail images by adopting the detection result to obtain detail images;
inputting the detail image into a bird recognition model for bird recognition to obtain a recognition result;
storing the identification result and the detail image into a database;
the bird detection model is obtained by training a deep learning neural network by taking image data of a plurality of bird position tags as first sample data;
the bird recognition model is obtained by training a neural network by taking image data of a plurality of bird-based tags as second sample data.
The further technical scheme is as follows: the bird detection model is obtained by training a deep learning neural network by taking image data of a plurality of bird position tags as first sample data, and comprises the following steps:
acquiring bird images;
labeling the bird position label on the bird image to obtain a first sample data set;
training a deep learning neural network using the first sample data set to obtain a bird detection model.
The further technical scheme is as follows: training the deep learning neural network with the first sample data set to obtain a bird detection model, comprising:
dividing the first sample data set into a first training set and a first test set;
setting parameters trained by a YOLOV4 algorithm;
inputting a first training set into a YOLOV4 algorithm to perform network model training so as to obtain a first initial model;
testing the first initial model by adopting a first test set to obtain a first test result;
judging whether the first test result meets the requirement or not;
if the first test result does not meet the requirements, executing the set parameters trained by the YOLOV4 algorithm;
and if the first test result meets the requirement, taking the first initial model as a bird detection model.
The further technical scheme is as follows: the bird recognition model is obtained by training a neural network by using image data of a plurality of bird-based tags as second sample data, and comprises the following steps:
acquiring detail images of birds;
labeling bird-based labels on the bird detail images to obtain a second sample data set;
the neural network is trained using the second sample dataset to obtain a bird recognition model.
The further technical scheme is as follows: the training of the neural network using the second sample dataset to obtain the bird recognition model comprises:
dividing the second sample data set into a second training set and a second test set;
setting parameters trained by a resnet50 algorithm;
inputting the second training set into a resnet50 algorithm to perform network model training so as to obtain a second initial model;
testing the second initial model by adopting a second test set to obtain a second test result;
judging whether the second test result meets the requirement or not;
if the second test result does not meet the requirements, executing the parameters for training the set resnet50 algorithm;
and if the second test result meets the requirement, taking the second initial model as a bird recognition model.
The further technical scheme is as follows: after the identification result and the detail image are stored in the database, the method further comprises the following steps:
analyzing the database and making a corresponding bird ecological environment treatment scheme to obtain a treatment plan, and feeding back the treatment plan to the terminal.
The invention also provides a bird recognition device, comprising:
the panoramic image acquisition unit is used for acquiring an image shot by the panoramic network camera to obtain a panoramic image;
the bird detection unit is used for inputting the panoramic image into a bird detection model to perform bird detection so as to obtain a detection result;
a detection judging unit for judging whether the detection result is a result of the existence of birds;
the segmentation unit is used for acquiring bird detail images from the ball machine if the detection result is a bird existence result, and segmenting the bird detail images by adopting the detection result to obtain detail images;
the identification unit is used for inputting the detail image into a bird identification model to carry out bird identification so as to obtain an identification result;
and the storage unit is used for storing the identification result and the detail image into a database.
The further technical scheme is as follows: further comprises:
the first construction unit is used for training the deep learning neural network by taking image data of a plurality of bird position tags as first sample data so as to obtain a bird detection model.
The invention also provides a computer device which comprises a memory and a processor, wherein the memory stores a computer program, and the processor realizes the method when executing the computer program.
The present invention also provides a storage medium storing a computer program which, when executed by a processor, performs the above-described method.
Compared with the prior art, the invention has the beneficial effects that: according to the invention, the panoramic network camera can shoot the monitored area without dead angles by acquiring the images shot by the panoramic network camera, so that the monitoring dead areas are not existed, the cost is low, the application site multipoint installation construction wiring is reduced, the panoramic images shot by shooting are input into the bird detection model for detection, the detail images shot by the linked dome camera are segmented and are input into the bird recognition model for recognition under the condition that birds exist, and the corresponding contents are saved into the database according to the recognition result, so that the bird is automatically recognized, the recognition efficiency and the accuracy are improved, the monitoring dead areas are not existed, and the cost is low.
The invention is further described below with reference to the drawings and specific embodiments.
Drawings
In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the drawings required for the description of the embodiments will be briefly described below, and it is obvious that the drawings in the following description are some embodiments of the present invention, and other drawings may be obtained according to these drawings without inventive effort for a person skilled in the art.
Fig. 1 is a schematic diagram of an application scenario of a bird recognition method according to an embodiment of the present invention;
FIG. 2 is a schematic flow chart of a bird identification method according to an embodiment of the present invention;
FIG. 3 is a schematic flow chart of a method for bird identification according to an embodiment of the present invention;
FIG. 4 is a schematic flow chart of a method for bird identification according to an embodiment of the present invention;
FIG. 5 is a schematic flow chart of a method for bird identification according to an embodiment of the present invention;
FIG. 6 is a schematic flow chart of a method for bird identification according to an embodiment of the present invention;
FIG. 7 is a flow chart of a method for bird identification according to another embodiment of the present invention;
FIG. 8 is a schematic block diagram of a bird recognition device provided by an embodiment of the present invention;
FIG. 9 is a schematic block diagram of a bird recognition device according to another embodiment of the present invention;
fig. 10 is a schematic block diagram of a computer device according to an embodiment of the present invention.
Detailed Description
The following description of the embodiments of the present invention will be made clearly and fully with reference to the accompanying drawings, in which it is evident that the embodiments described are some, but not all embodiments of the invention. All other embodiments, which can be made by those skilled in the art based on the embodiments of the invention without making any inventive effort, are intended to be within the scope of the invention.
It should be understood that the terms "comprises" and "comprising," when used in this specification and the appended claims, specify the presence of stated features, integers, steps, operations, elements, and/or components, but do not preclude the presence or addition of one or more other features, integers, steps, operations, elements, components, and/or groups thereof.
It is also to be understood that the terminology used in the description of the invention herein is for the purpose of describing particular embodiments only and is not intended to be limiting of the invention. As used in this specification and the appended claims, the singular forms "a," "an," and "the" are intended to include the plural forms as well, unless the context clearly indicates otherwise.
It should be further understood that the term "and/or" as used in the present specification and the appended claims refers to any and all possible combinations of one or more of the associated listed items, and includes such combinations.
Referring to fig. 1 and fig. 2, fig. 1 is a schematic diagram of an application scenario of a bird recognition method according to an embodiment of the present invention. Fig. 2 is a schematic flow chart of a bird identification method according to an embodiment of the present invention. The bird identification method is applied to the server. After the panoramic image is acquired through the panoramic network camera, whether birds exist or not is detected by the server, when the birds exist, the detection result is that coordinates of a rectangular frame where the birds exist, namely position information of the birds, are output, after that, the server acquires detail images of the birds from the dome, only comprises area images with only bird details after dividing the detail images according to the position information, the area images are input into a bird recognition model for recognition, so that birds are stored in a database for the terminal to fetch and review.
Fig. 2 is a schematic flow chart of a bird identification method according to an embodiment of the present invention. As shown in fig. 2, the method includes the following steps S110 to S160.
S110, acquiring an image shot by the panoramic network camera to obtain a panoramic image.
In the present embodiment, the panoramic image is an image taken from a panoramic network camera after the panoramic network camera is arranged in the monitored area.
The panoramic network camera is used for covering the whole range to be monitored, no monitoring blind area exists, the panoramic network camera is used for monitoring the application site in 360-degree dead angle-free range, one panoramic network camera is used for fully covering the whole monitoring area, detection and identification of types of birds are realized, detection of the blind area is realized, multi-point installation construction wiring of the application site is reduced, and hardware cost of a plurality of cameras is reduced.
S120, inputting the panoramic image into a bird detection model for bird detection so as to obtain a detection result.
In this embodiment, the detection result refers to whether or not there is a bird present category, and when the detected bird present category, the detection result also includes a rectangular frame where the bird is located, that is, a location where the bird is located.
The bird detection model is obtained by training a deep learning neural network by taking image data of a plurality of bird position tags as first sample data.
In an embodiment, referring to fig. 3, the bird detection model is obtained by training a deep learning neural network by using image data of a plurality of bird-site tags as first sample data, and includes steps S121 to S123.
S121, acquiring bird images.
In the present embodiment, bird images refer to a related picture of birds collected from a network and a picture collected in an actual environment by assuming a panoramic network camera, and these bird images include images that contain no birds or have contained birds.
S122, marking the bird position label on the bird image to obtain a first sample data set.
In this embodiment, the first sample data set refers to data that is tagged with bird-location and that can be used to train a deep learning neural network.
Bird images are collected and manually shot on a network, bird images under various environmental backgrounds are collected as much as possible, and marked and stored as xml data format files. The marked bird images are randomly divided into a first training set and a first testing set according to the proportion of 9:1, and of course, the first sample data set can be divided according to different proportions according to actual requirements.
Specifically, a labeling tool labelImg is used for carrying out coordinate labeling on bird images, if the bird images exist, a rectangular bird frame is labeled, so that the bird is just framed optimally, the bird is labeled as a bird tag and a position information tag, and if the bird images exist, the bird images are labeled as blank tags, so that the deep learning neural network is trained.
S123, training the deep learning neural network by using the first sample data set to obtain a bird detection model.
In the present embodiment, the bird detection model refers to a model that has been trained and can be used to directly perform bird detection on an input image to obtain positional information of whether birds are present or not and when birds are present.
In one embodiment, referring to fig. 4, the step S123 may include steps S1231 to S1236.
S1231, dividing the first sample data set into a first training set and a first test set.
In this embodiment, the first training set is image data for training YOLOV4 algorithm; the first test set is image data for testing the trained YOLOV4 algorithm.
S1232, setting parameters trained by a YOLOV4 algorithm;
s1233, inputting the first training set into a YOLOV4 algorithm to perform network model training so as to obtain a first initial model.
In this embodiment, the first initial model is trained by using YOLOV4 algorithm in the deep learning neural network, training by adopting a random gradient descent algorithm, stopping training when the loss function value is reduced to be stable, namely when the loss value of the loss function is stable, and storing the model obtained by training.
And (3) taking the pre-training data model with an open source as an initialized parameter, stopping training when the loss function value is reduced to a stable area and is at a point smaller than 1.0, and outputting a first initial model.
S1234, testing the first initial model by adopting the first test set to obtain a first test result.
In this embodiment, the first test result refers to a result obtained by testing the first initial model with the first test set.
S1235, judging whether the first test result meets the requirement;
if the first test result does not meet the requirement, executing the step S1232;
s1236, if the first test result meets the requirement, the first initial model is used as a bird detection model.
Specifically, the mAP (mean average precision ) index is used to determine the test result of the test set on the initial model, and if mAP is less than 0.95, the training parameter setting is modified or the data set is added for retraining until the requirement that mAP is greater than 0.95 is met. The mAP index is the average accuracy over one category, and then the average accuracy over all categories is further averaged.
Training by using a deep learning neural network, training by adopting a random gradient descent algorithm, testing and evaluating a first initial model when the function value with loss is reduced to tend to be stable, and selecting an optimal model, thereby forming the bird detection model.
S130, judging whether the detection result is a result of birds;
if no birds exist as a result of the detection, step S110 is performed.
And S140, if the detection result is that birds exist, acquiring bird detail images from the ball machine, and dividing the bird detail images by adopting the detection result to obtain detail images.
In this embodiment, the bird detail image refers to an image of birds from a ball machine.
And according to the detected coordinate position, linking the ball machine to acquire a bird image with clearer bird detail information and outputting the segmented detail image.
The panoramic network camera is used for detecting the bird-linked dome camera to output images with clear local details, so that powerful guarantee is provided for bird identification.
And S150, inputting the detail image into a bird recognition model to perform bird recognition so as to obtain a recognition result.
In this embodiment, the recognition result refers to the category to which birds in the detail image belong.
The bird recognition model is obtained by training a neural network through image data of a plurality of bird-based tags as second sample data.
In an embodiment, referring to fig. 5, the bird recognition model is obtained by training the neural network by using image data of a plurality of bird-based tags as the second sample data, and includes steps S151 to S153.
S151, acquiring detail images of birds.
In this embodiment, the detailed images of birds refer to acquiring images of different kinds of birds by adopting a network-collected image and a network-open-source image library.
S152, labeling bird-based labels on the bird-based detail images to obtain a second sample data set.
In this embodiment, the second sample data set refers to data that is labeled with an avian species tag and that can be used to train a neural network.
The data for identifying the bird species are marked by numbers as labels, and if the birds are N species, the birds are respectively marked as 0,1,2,3, … and N-1.
And S153, training the neural network by using the second sample data set to obtain a bird recognition model.
In the present embodiment, the bird recognition model refers to a model that has been trained and can be used for bird species recognition directly on an input image.
In one embodiment, referring to fig. 6, the step S153 may include steps S1531 to S1536.
S1531, dividing the second sample data set into a second training set and a second test set.
In this embodiment, the first training set is image data for training a resnet50 algorithm; the first test set is image data for testing the trained resnet50 algorithm.
S1532, setting parameters trained by a resnet50 algorithm;
s1533, inputting the second training set into a resnet50 algorithm for training the network model to obtain a second initial model.
In this embodiment, the second initial model is trained by using a resnet50 algorithm in the deep learning neural network, and when the loss function value is reduced to be stable, that is, when the loss value of the loss function is stable, the training is stopped, and the model obtained by training is stored.
Specifically, the network model is output after the network model parameters tend to be stable and the loss error is less than 1.0, and the training is stopped by adopting the resnet50 algorithm to perform bird species recognition network training and also taking the open-source pre-training model as the eating actual speech parameters of the training model.
S1534, testing the second initial model by adopting the second test set to obtain a second test result.
In this embodiment, the second test result refers to a result obtained by testing the second initial model with the second test set.
S1535, judging whether the second test result meets the requirement;
if the second test result does not meet the requirement, executing the step S1532;
s1536, if the second test result meets the requirement, using the second initial model as a bird recognition model.
S160, storing the identification result and the detail image into a database;
specifically, a database is established for the recognition result and the detail image, and the image and the result obtained by the subsequent recognition are stored in the database.
If birds are stored in the database, the birds are compiled into the bird database, and if the birds are not known, the bird pictures are saved, so that a new bird sample is provided for further expanding the bird recognition system
According to the bird recognition method, the image obtained by shooting the panoramic network camera is obtained, the panoramic network camera can shoot a monitored area without dead angles, no monitoring dead area exists, the application site multipoint installation construction wiring is reduced, the cost is low, the panoramic image obtained by shooting is input into the bird detection model for detection, under the condition that birds exist, the detail image is shot by the linked dome camera, the detail image is segmented and then is input into the bird recognition model for recognition, the corresponding content is saved into the database according to the recognition result, the automatic recognition of birds is realized, the recognition efficiency and accuracy are improved, the monitoring dead area does not exist, and the cost is low.
Fig. 7 is a flow chart of a bird identification method according to another embodiment of the present invention. As shown in fig. 7, the bird recognition method of the present embodiment includes steps S210 to S270. Steps S210 to S260 are similar to steps S110 to S160 in the above embodiment, and are not described herein. Step S270 added in the present embodiment is described in detail below.
S270, analyzing the database and making a corresponding bird ecological environment treatment scheme to obtain a treatment plan, and feeding back the treatment plan to the terminal.
In this embodiment, the governance plan refers to a related bird ecological environment governance plan established according to bird recognition results of a specific monitoring area such as an airport.
Through the analysis of the database, a corresponding bird ecological environment treatment plan is formulated, and the plan is sent to airport personnel to be executed by the airport personnel, so that the damage to birds can be reduced to the greatest extent, and a good bird repelling effect can be achieved.
Fig. 8 is a schematic block diagram of a bird recognition device 300 provided in an embodiment of the present invention. As shown in fig. 8, the present invention also provides a bird recognition device 300 corresponding to the above bird recognition method. The bird recognition apparatus 300 includes a unit for performing the above bird recognition method, and the apparatus may be configured in a server. Specifically, referring to fig. 8, the bird recognition apparatus 300 includes a panoramic image acquisition unit 301, a bird detection unit 302, a detection judgment unit 303, a division unit 304, a recognition unit 305, and a storage unit 306.
A panoramic image obtaining unit 301, configured to obtain an image obtained by capturing by a panoramic webcam, so as to obtain a panoramic image; a bird detection unit 302, configured to input the panoramic image into a bird detection model for bird detection, so as to obtain a detection result; a detection judgment unit 303 for judging whether the detection result is a result of the presence of birds; the segmentation unit 304 is configured to acquire a bird detail image from the dome camera if the detection result is a bird existence result, and segment the bird detail image by using the detection result to obtain a detail image; a recognition unit 305, configured to input the detail image into a bird recognition model for bird recognition, so as to obtain a recognition result; and the storage unit 306 is used for storing the identification result and the detail image into a database.
In one embodiment, the bird recognition device 300 further includes a first construction unit.
The first construction unit is used for training the deep learning neural network by taking image data of a plurality of bird position tags as first sample data so as to obtain a bird detection model.
In an embodiment, the first construction unit includes an avian image acquisition subunit, a position tag labeling subunit, and a first training subunit.
A bird image acquisition subunit configured to acquire a bird image; the position tag labeling subunit is used for labeling the bird position tags on the bird images so as to obtain a first sample data set; and the first training subunit is used for training the deep learning neural network by adopting the first sample data set so as to obtain the bird detection model.
In an embodiment, the first training subunit includes a first dividing module, a first setting module, a first model obtaining module, a first testing module, and a first judging module.
The first dividing module is used for dividing the first sample data set into a first training set and a first testing set; the first setting module is used for setting parameters trained by the YOLOV4 algorithm; the first model acquisition module is used for inputting a first training set into a YOLOV4 algorithm to perform network model training so as to obtain a first initial model; the first test module is used for testing the first initial model by adopting a first test set so as to obtain a first test result; the first judging module is used for judging whether the first test result meets the requirement or not; if the first test result does not meet the requirements, executing the set parameters trained by the YOLOV4 algorithm; and if the first test result meets the requirement, taking the first initial model as a bird detection model.
In one embodiment, the bird recognition device 300 further includes a second construction unit.
And the second construction unit is used for training the neural network by taking the image data of the plurality of bird-based tags as second sample data so as to obtain a bird-based model.
In an embodiment, the second building unit includes a detail image acquiring subunit, a category label labeling subunit, and a second training subunit.
A detail image acquisition subunit, configured to acquire a detail image of the bird; the class label labeling subunit is used for labeling the bird class labels on the detailed images of the birds so as to obtain a second sample data set; and the second training subunit is used for training the neural network by adopting the second sample data set so as to obtain the bird recognition model.
In an embodiment, the second training subunit includes a second dividing module, a second setting module, a second model obtaining module, a second testing module, and a second judging module.
The second dividing module is used for dividing the second sample data set into a second training set and a second testing set; the second setting module is used for setting parameters trained by the resnet50 algorithm; the second model acquisition module is used for inputting a second training set into a resnet50 algorithm to perform network model training so as to obtain a second initial model; the second testing module is used for testing the second initial model by adopting a second testing set so as to obtain a second testing result; the second judging module is used for judging whether the second test result meets the requirements; if the second test result does not meet the requirements, executing the parameters for training the set resnet50 algorithm; and if the second test result meets the requirement, taking the second initial model as a bird recognition model.
Fig. 9 is a schematic block diagram of a bird recognition device 300 according to another embodiment of the present invention. As shown in fig. 9, the bird recognition device 300 of the present embodiment is an addition of the formulation unit 307 to the above-described embodiment.
And the making unit 307 is used for analyzing the database and making a corresponding bird ecological environment treatment scheme so as to obtain a treatment plan, and feeding back the treatment plan to the terminal.
It should be noted that, as those skilled in the art can clearly understand, the specific implementation process of the bird recognition device 300 and each unit may refer to the corresponding description in the foregoing method embodiment, and for convenience and brevity of description, the description is omitted here.
The bird recognition device 300 described above may be implemented in the form of a computer program that can run on a computer apparatus as shown in fig. 10.
Referring to fig. 10, fig. 10 is a schematic block diagram of a computer device according to an embodiment of the present application. The computer device 500 may be a server, which may be a stand-alone server or a server cluster formed by a plurality of servers.
With reference to FIG. 10, the computer device 500 includes a processor 502, memory, and a network interface 505 connected by a system bus 501, where the memory may include a non-volatile storage medium 503 and an internal memory 504.
The non-volatile storage medium 503 may store an operating system 5031 and a computer program 5032. The computer program 5032 includes program instructions that, when executed, cause the processor 502 to perform a bird identification method.
The processor 502 is used to provide computing and control capabilities to support the operation of the overall computer device 500.
The internal memory 504 provides an environment for the execution of a computer program 5032 in the non-volatile storage medium 503, which computer program 5032, when executed by the processor 502, causes the processor 502 to perform a bird identification method.
The network interface 505 is used for network communication with other devices. Those skilled in the art will appreciate that the architecture shown in fig. 10 is merely a block diagram of a portion of the architecture in connection with the present application and is not intended to limit the computer device 500 to which the present application is applied, and that a particular computer device 500 may include more or fewer components than shown, or may combine certain components, or have a different arrangement of components.
Wherein the processor 502 is configured to execute a computer program 5032 stored in a memory to implement the steps of:
acquiring an image shot by a panoramic network camera to obtain a panoramic image; inputting the panoramic image into a bird detection model for bird detection to obtain a detection result; judging whether the detection result is a result of birds; if the detection result is that birds exist, acquiring bird detail images from the ball machine, and dividing the bird detail images by adopting the detection result to obtain detail images; inputting the detail image into a bird recognition model for bird recognition to obtain a recognition result; and storing the identification result and the detail image into a database.
The bird detection model is obtained by training a deep learning neural network by taking image data of a plurality of bird position tags as first sample data; the bird recognition model is obtained by training a neural network by taking image data of a plurality of bird-based tags as second sample data.
In one embodiment, when implementing the bird detection model by training the deep learning neural network using the image data of the plurality of bird-location tags as the first sample data, the processor 502 specifically implements the following steps:
acquiring bird images; labeling the bird position label on the bird image to obtain a first sample data set; training a deep learning neural network using the first sample data set to obtain a bird detection model.
In one embodiment, the processor 502 implements the training the deep learning neural network using the first sample data set to obtain the bird detection model, and specifically implements the following steps:
dividing the first sample data set into a first training set and a first test set; setting parameters trained by a YOLOV4 algorithm; inputting a first training set into a YOLOV4 algorithm to perform network model training so as to obtain a first initial model; testing the first initial model by adopting a first test set to obtain a first test result; judging whether the first test result meets the requirement or not; if the first test result does not meet the requirements, executing the set parameters trained by the YOLOV4 algorithm; and if the first test result meets the requirement, taking the first initial model as a bird detection model.
In one embodiment, when implementing the bird recognition model by training the neural network using the image data of the plurality of bird-like tags as the second sample data, the processor 502 specifically implements the following steps:
acquiring detail images of birds; labeling bird-based labels on the bird detail images to obtain a second sample data set; the neural network is trained using the second sample dataset to obtain a bird recognition model.
In one embodiment, the processor 502 performs the following steps when performing the training of the neural network using the second sample data set to obtain the bird recognition model:
dividing the second sample data set into a second training set and a second test set; setting parameters trained by a resnet50 algorithm; inputting the second training set into a resnet50 algorithm to perform network model training so as to obtain a second initial model; testing the second initial model by adopting a second test set to obtain a second test result; judging whether the second test result meets the requirement or not; if the second test result does not meet the requirements, executing the parameters for training the set resnet50 algorithm; and if the second test result meets the requirement, taking the second initial model as a bird recognition model.
In one embodiment, after the step of saving the recognition result and the detailed image in the database is performed, the processor 502 further performs the following steps:
analyzing the database and making a corresponding bird ecological environment treatment scheme to obtain a treatment plan, and feeding back the treatment plan to the terminal.
It should be appreciated that in embodiments of the present application, the processor 502 may be a central processing unit (Central Processing Unit, CPU), the processor 502 may also be other general purpose processors, digital signal processors (Digital Signal Processor, DSPs), application specific integrated circuits (Application Specific Integrated Circuit, ASICs), off-the-shelf programmable gate arrays (Field-Programmable Gate Array, FPGAs) or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, or the like. Wherein the general purpose processor may be a microprocessor or the processor may be any conventional processor or the like.
Those skilled in the art will appreciate that all or part of the flow in a method embodying the above described embodiments may be accomplished by computer programs instructing the relevant hardware. The computer program comprises program instructions, and the computer program can be stored in a storage medium, which is a computer readable storage medium. The program instructions are executed by at least one processor in the computer system to implement the flow steps of the embodiments of the method described above.
Accordingly, the present invention also provides a storage medium. The storage medium may be a computer readable storage medium. The storage medium stores a computer program which, when executed by a processor, causes the processor to perform the steps of:
acquiring an image shot by a panoramic network camera to obtain a panoramic image; inputting the panoramic image into a bird detection model for bird detection to obtain a detection result; judging whether the detection result is a result of birds; if the detection result is that birds exist, acquiring bird detail images from the ball machine, and dividing the bird detail images by adopting the detection result to obtain detail images; inputting the detail image into a bird recognition model for bird recognition to obtain a recognition result; and storing the identification result and the detail image into a database.
The bird detection model is obtained by training a deep learning neural network by taking image data of a plurality of bird position tags as first sample data; the bird recognition model is obtained by training a neural network by taking image data of a plurality of bird-based tags as second sample data.
In one embodiment, when the processor executes the computer program to implement the bird detection model as a step of training the deep learning neural network by using image data of a plurality of bird-location tags as first sample data, the processor specifically implements the following steps:
acquiring bird images; labeling the bird position label on the bird image to obtain a first sample data set; training a deep learning neural network using the first sample data set to obtain a bird detection model.
In one embodiment, the processor, when executing the computer program to implement the training of the deep learning neural network using the first sample data set to obtain the bird detection model step, specifically implements the following steps:
dividing the first sample data set into a first training set and a first test set; setting parameters trained by a YOLOV4 algorithm; inputting a first training set into a YOLOV4 algorithm to perform network model training so as to obtain a first initial model; testing the first initial model by adopting a first test set to obtain a first test result; judging whether the first test result meets the requirement or not; if the first test result does not meet the requirements, executing the set parameters trained by the YOLOV4 algorithm; and if the first test result meets the requirement, taking the first initial model as a bird detection model.
In one embodiment, when the processor executes the computer program to implement the bird recognition model as a step of training the neural network by using image data of a plurality of bird-like tags as the second sample data, the method specifically includes the steps of:
acquiring detail images of birds; labeling bird-based labels on the bird detail images to obtain a second sample data set; the neural network is trained using the second sample dataset to obtain a bird recognition model.
In one embodiment, the processor, when executing the computer program to implement the training neural network using the second sample data set to obtain the bird recognition model step, specifically implements the following steps:
dividing the second sample data set into a second training set and a second test set; setting parameters trained by a resnet50 algorithm; inputting the second training set into a resnet50 algorithm to perform network model training so as to obtain a second initial model; testing the second initial model by adopting a second test set to obtain a second test result; judging whether the second test result meets the requirement or not; if the second test result does not meet the requirements, executing the parameters for training the set resnet50 algorithm; and if the second test result meets the requirement, taking the second initial model as a bird recognition model.
In one embodiment, after executing the computer program to implement the step of saving the recognition result and the detailed image into a database, the processor further implements the steps of:
analyzing the database and making a corresponding bird ecological environment treatment scheme to obtain a treatment plan, and feeding back the treatment plan to the terminal.
The storage medium may be a U-disk, a removable hard disk, a Read-Only Memory (ROM), a magnetic disk, or an optical disk, or other various computer-readable storage media that can store program codes.
Those of ordinary skill in the art will appreciate that the elements and algorithm steps described in connection with the embodiments disclosed herein may be embodied in electronic hardware, in computer software, or in a combination of the two, and that the elements and steps of the examples have been generally described in terms of function in the foregoing description to clearly illustrate the interchangeability of hardware and software. Whether such functionality is implemented as hardware or software depends upon the particular application and design constraints imposed on the solution. Skilled artisans may implement the described functionality in varying ways for each particular application, but such implementation decisions should not be interpreted as causing a departure from the scope of the present invention.
In the several embodiments provided by the present invention, it should be understood that the disclosed apparatus and method may be implemented in other manners. For example, the device embodiments described above are merely illustrative. For example, the division of each unit is only one logic function division, and there may be another division manner in actual implementation. For example, multiple units or components may be combined or may be integrated into another system, or some features may be omitted, or not performed.
The steps in the method of the embodiment of the invention can be sequentially adjusted, combined and deleted according to actual needs. The units in the device of the embodiment of the invention can be combined, divided and deleted according to actual needs. In addition, each functional unit in the embodiments of the present invention may be integrated in one processing unit, or each unit may exist alone physically, or two or more units may be integrated in one unit.
The integrated unit may be stored in a storage medium if implemented in the form of a software functional unit and sold or used as a stand-alone product. Based on such understanding, the technical solution of the present invention is essentially or a part contributing to the prior art, or all or part of the technical solution may be embodied in the form of a software product stored in a storage medium, comprising several instructions for causing a computer device (which may be a personal computer, a terminal, a network device, etc.) to perform all or part of the steps of the method according to the embodiments of the present invention.
While the invention has been described with reference to certain preferred embodiments, it will be understood by those skilled in the art that various changes and substitutions of equivalents may be made and equivalents will be apparent to those skilled in the art without departing from the scope of the invention. Therefore, the protection scope of the invention is subject to the protection scope of the claims.
Claims (5)
1. A method of bird identification comprising:
acquiring an image shot by a panoramic network camera to obtain a panoramic image;
inputting the panoramic image into a bird detection model for bird detection to obtain a detection result;
judging whether the detection result is a result of birds;
if the detection result is that birds exist, acquiring bird detail images from the ball machine, and dividing the bird detail images by adopting the detection result to obtain detail images;
inputting the detail image into a bird recognition model for bird recognition to obtain a recognition result;
storing the identification result and the detail image into a database;
the bird detection model is obtained by training a deep learning neural network by taking image data of a plurality of bird position tags as first sample data;
The bird recognition model is obtained by training a neural network by taking image data of a plurality of bird-based tags as second sample data;
the bird detection model is obtained by training a deep learning neural network by taking image data of a plurality of bird position tags as first sample data, and comprises the following steps:
acquiring bird images;
labeling the bird position label on the bird image to obtain a first sample data set;
training a deep learning neural network by adopting a first sample data set to obtain a bird detection model;
training the deep learning neural network with the first sample data set to obtain a bird detection model, comprising:
dividing the first sample data set into a first training set and a first test set;
setting parameters trained by a YOLOV4 algorithm;
inputting a first training set into a YOLOV4 algorithm to perform network model training so as to obtain a first initial model;
testing the first initial model by adopting a first test set to obtain a first test result;
judging whether the first test result meets the requirement or not;
if the first test result does not meet the requirements, executing the set parameters trained by the YOLOV4 algorithm;
If the first test result meets the requirement, the first initial model is used as a bird detection model;
the bird recognition model is obtained by training a neural network by using image data of a plurality of bird-based tags as second sample data, and comprises the following steps:
acquiring detail images of birds;
labeling bird-based labels on the bird detail images to obtain a second sample data set;
training a neural network by adopting a second sample data set to obtain a bird recognition model;
the training of the neural network using the second sample dataset to obtain the bird recognition model comprises:
dividing the second sample data set into a second training set and a second test set;
setting parameters trained by a resnet50 algorithm;
inputting the second training set into a resnet50 algorithm to perform network model training so as to obtain a second initial model;
testing the second initial model by adopting a second test set to obtain a second test result;
judging whether the second test result meets the requirement or not;
if the second test result does not meet the requirements, executing the parameters for training the set resnet50 algorithm;
and if the second test result meets the requirement, taking the second initial model as a bird recognition model.
2. The bird recognition method of claim 1, wherein after storing the recognition result and the detail image in a database, further comprising:
analyzing the database and making a corresponding bird ecological environment treatment scheme to obtain a treatment plan, and feeding back the treatment plan to the terminal.
3. Bird recognition device, characterized by comprising:
the panoramic image acquisition unit is used for acquiring an image shot by the panoramic network camera to obtain a panoramic image;
the bird detection unit is used for inputting the panoramic image into a bird detection model to perform bird detection so as to obtain a detection result;
a detection judging unit for judging whether the detection result is a result of the existence of birds;
the segmentation unit is used for acquiring bird detail images from the ball machine if the detection result is a bird existence result, and segmenting the bird detail images by adopting the detection result to obtain detail images;
the identification unit is used for inputting the detail image into a bird identification model to carry out bird identification so as to obtain an identification result;
the storage unit is used for storing the identification result and the detail image into a database;
Further comprises:
the first construction unit is used for training the deep learning neural network by taking image data of a plurality of bird position tags as first sample data so as to obtain a bird detection model;
the first construction unit comprises a bird image acquisition subunit, a position tag labeling subunit and a first training subunit;
a bird image acquisition subunit configured to acquire a bird image; the position tag labeling subunit is used for labeling the bird position tags on the bird images so as to obtain a first sample data set; the first training subunit is used for training the deep learning neural network by adopting the first sample data set so as to obtain a bird detection model;
the first training subunit comprises a first dividing module, a first setting module, a first model acquisition module, a first testing module and a first judging module;
the first dividing module is used for dividing the first sample data set into a first training set and a first testing set; the first setting module is used for setting parameters trained by the YOLOV4 algorithm; the first model acquisition module is used for inputting a first training set into a YOLOV4 algorithm to perform network model training so as to obtain a first initial model; the first test module is used for testing the first initial model by adopting a first test set so as to obtain a first test result; the first judging module is used for judging whether the first test result meets the requirement or not; if the first test result does not meet the requirements, executing the set parameters trained by the YOLOV4 algorithm; if the first test result meets the requirement, the first initial model is used as a bird detection model;
The bird recognition device further includes a second construction unit;
training a neural network by using image data of a plurality of bird-based tags as second sample data to obtain a bird recognition model;
the second construction unit comprises a detail image acquisition subunit, a category label marking subunit and a second training subunit;
a detail image acquisition subunit, configured to acquire a detail image of the bird; the class label labeling subunit is used for labeling the bird class labels on the detailed images of the birds so as to obtain a second sample data set; a second training subunit for training the neural network with the second sample dataset to obtain a bird recognition model;
the second training subunit comprises a second dividing module, a second setting module, a second model acquisition module, a second testing module and a second judging module;
the second dividing module is used for dividing the second sample data set into a second training set and a second testing set; the second setting module is used for setting parameters trained by the resnet50 algorithm; the second model acquisition module is used for inputting a second training set into a resnet50 algorithm to perform network model training so as to obtain a second initial model; the second testing module is used for testing the second initial model by adopting a second testing set so as to obtain a second testing result; the second judging module is used for judging whether the second test result meets the requirements; if the second test result does not meet the requirements, executing the parameters for training the set resnet50 algorithm; and if the second test result meets the requirement, taking the second initial model as a bird recognition model.
4. A computer device, characterized in that it comprises a memory on which a computer program is stored and a processor which, when executing the computer program, implements the method according to any of claims 1-2.
5. A storage medium storing a computer program which, when executed by a processor, performs the method of any one of claims 1 to 2.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202010557975.XA CN111709421B (en) | 2020-06-18 | 2020-06-18 | Bird identification method, bird identification device, computer equipment and storage medium |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202010557975.XA CN111709421B (en) | 2020-06-18 | 2020-06-18 | Bird identification method, bird identification device, computer equipment and storage medium |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN111709421A CN111709421A (en) | 2020-09-25 |
| CN111709421B true CN111709421B (en) | 2023-06-27 |
Family
ID=72541434
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202010557975.XA Active CN111709421B (en) | 2020-06-18 | 2020-06-18 | Bird identification method, bird identification device, computer equipment and storage medium |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN111709421B (en) |
Families Citing this family (7)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN112541432A (en) * | 2020-12-11 | 2021-03-23 | 上海品览数据科技有限公司 | Video livestock identity authentication system and method based on deep learning |
| CN112668444A (en) * | 2020-12-24 | 2021-04-16 | 南京泓图人工智能技术研究院有限公司 | Bird detection and identification method based on YOLOv5 |
| CN113076860B (en) * | 2021-03-30 | 2022-02-25 | 南京大学环境规划设计研究院集团股份公司 | Bird detection system under field scene |
| CN113344206A (en) * | 2021-06-25 | 2021-09-03 | 江苏大学 | Knowledge distillation method, device and equipment integrating channel and relation feature learning |
| CN115100687A (en) * | 2022-07-18 | 2022-09-23 | 中国科学院半导体研究所 | Bird detection method and device in ecological region and electronic equipment |
| CN115690448A (en) * | 2022-11-09 | 2023-02-03 | 广东省科学院动物研究所 | AI-based bird species identification method and device |
| CN118097721B (en) * | 2024-04-29 | 2024-06-25 | 江西师范大学 | Wetland bird recognition method and system based on multi-source remote sensing observation and deep learning |
Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN107087144A (en) * | 2017-05-13 | 2017-08-22 | 西安费斯达自动化工程有限公司 | Panorama and precise image/spherical crown variable excitation chirm integrative detection system |
| CN109033975A (en) * | 2018-06-27 | 2018-12-18 | 山东大学 | Birds detection, identification and method for tracing and device in a kind of monitoring of seashore |
| CN110059641A (en) * | 2019-04-23 | 2019-07-26 | 重庆工商大学 | Depth birds recognizer based on more preset points |
| CN110889841A (en) * | 2019-11-28 | 2020-03-17 | 江苏电力信息技术有限公司 | YOLOv 3-based bird detection algorithm for power transmission line |
| CN110969107A (en) * | 2019-11-25 | 2020-04-07 | 上海交通大学 | Bird population identification analysis method and system based on network model |
-
2020
- 2020-06-18 CN CN202010557975.XA patent/CN111709421B/en active Active
Patent Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN107087144A (en) * | 2017-05-13 | 2017-08-22 | 西安费斯达自动化工程有限公司 | Panorama and precise image/spherical crown variable excitation chirm integrative detection system |
| CN109033975A (en) * | 2018-06-27 | 2018-12-18 | 山东大学 | Birds detection, identification and method for tracing and device in a kind of monitoring of seashore |
| CN110059641A (en) * | 2019-04-23 | 2019-07-26 | 重庆工商大学 | Depth birds recognizer based on more preset points |
| CN110969107A (en) * | 2019-11-25 | 2020-04-07 | 上海交通大学 | Bird population identification analysis method and system based on network model |
| CN110889841A (en) * | 2019-11-28 | 2020-03-17 | 江苏电力信息技术有限公司 | YOLOv 3-based bird detection algorithm for power transmission line |
Non-Patent Citations (2)
| Title |
|---|
| Seolhee Lee，etc.Bird Detection in Agriculture Environment using Image Processing and Neural Network.《2019 6th International Conference on Control, Decision and Information Technologies (CoDIT)》.2019,全文. * |
| 刘坚.基于深度神经网络的鸟类图像识别系统设计.《农业装备与车辆工程》.2019,全文. * |
Also Published As
| Publication number | Publication date |
|---|---|
| CN111709421A (en) | 2020-09-25 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN111709421B (en) | Bird identification method, bird identification device, computer equipment and storage medium | |
| WO2021047232A1 (en) | Interaction behavior recognition method, apparatus, computer device, and storage medium | |
| CN108197658B (en) | Image annotation information processing method, device, server and system | |
| Yang et al. | Deformable convolution and coordinate attention for fast cattle detection | |
| French et al. | Deep neural networks for analysis of fisheries surveillance video and automated monitoring of fish discards | |
| CN109886928B (en) | Target cell marking method, device, storage medium and terminal equipment | |
| CN111709374B (en) | Bird condition detection method, bird condition detection device, computer equipment and storage medium | |
| CN111598164B (en) | Method, device, electronic equipment and storage medium for identifying attribute of target object | |
| CN110363138A (en) | Model training method, image processing method, device, terminal and storage medium | |
| CN107680080B (en) | Sample library establishing method and checking method for livestock, storage medium and electronic equipment | |
| CN110163798B (en) | Method and system for detecting damage of purse net in fishing ground | |
| CN110307903B (en) | Method for dynamically measuring non-contact temperature of specific part of poultry | |
| CN113439227A (en) | Capturing and storing magnified images | |
| CN111325217A (en) | Data processing method, device, system and medium | |
| CN111325181A (en) | State monitoring method and device, electronic equipment and storage medium | |
| Vilas et al. | Use of computer vision onboard fishing vessels to quantify catches: The iObserver | |
| CN110210314B (en) | Face detection method, device, computer equipment and storage medium | |
| CN111553184A (en) | Small target detection method and device based on electronic purse net and electronic equipment | |
| Hasan et al. | Framework for fish freshness detection and rotten fish removal in Bangladesh using mask R–CNN method with robotic arm and fisheye analysis | |
| CN111667450A (en) | Ship quantity counting method and device and electronic equipment | |
| CN111582012A (en) | Method and device for detecting small target ship | |
| CN115661717A (en) | Livestock crawling behavior marking method and device, electronic equipment and storage medium | |
| CN114973022A (en) | Animal husbandry livestock monitoring and extracting method based on high spatial resolution remote sensing image | |
| Heck et al. | Defining Quality Requirements for a Trustworthy AI Wildflower Monitoring Platform | |
| CN109726741B (en) | Method and device for detecting multiple target objects |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |