CN108197591A - A kind of birds individual discrimination method based on multiple features fusion transfer learning - Google Patents
A kind of birds individual discrimination method based on multiple features fusion transfer learning Download PDFInfo
- Publication number
- CN108197591A CN108197591A CN201810058223.1A CN201810058223A CN108197591A CN 108197591 A CN108197591 A CN 108197591A CN 201810058223 A CN201810058223 A CN 201810058223A CN 108197591 A CN108197591 A CN 108197591A
- Authority
- CN
- China
- Prior art keywords
- birds
- chirping
- feature vector
- feature
- identification model
- 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.)
- Pending
Links
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F2218/00—Aspects of pattern recognition specially adapted for signal processing
- G06F2218/12—Classification; Matching
-
- 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
- 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
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F2218/00—Aspects of pattern recognition specially adapted for signal processing
- G06F2218/02—Preprocessing
Landscapes
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Physics & Mathematics (AREA)
- Artificial Intelligence (AREA)
- General Physics & Mathematics (AREA)
- General Engineering & Computer Science (AREA)
- Data Mining & Analysis (AREA)
- Evolutionary Computation (AREA)
- Life Sciences & Earth Sciences (AREA)
- Computing Systems (AREA)
- Software Systems (AREA)
- Molecular Biology (AREA)
- Computational Linguistics (AREA)
- Biophysics (AREA)
- Biomedical Technology (AREA)
- Mathematical Physics (AREA)
- General Health & Medical Sciences (AREA)
- Health & Medical Sciences (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Signal Processing (AREA)
- Bioinformatics & Cheminformatics (AREA)
- Bioinformatics & Computational Biology (AREA)
- Evolutionary Biology (AREA)
- Catching Or Destruction (AREA)
- Image Analysis (AREA)
Abstract
The invention discloses a kind of birds individual discrimination method based on multiple features fusion transfer learning, including：S1, known chirping of birds signal is pre-processed；S2, framing and windowing process are carried out to pretreated chirping of birds signal, the chirping of birds signal grown to obtained framing generates sonagram using chirplet；S3, depth convolutional neural networks are handled；S4, the sonagram generated in S2 is input to the feature vector that different layers are obtained in the neural network of S3, final feature vector is obtained after fusion；S5, final feature vector is input in support vector machines, identification model is obtained after training；S6, performance detection is carried out to obtained identification model, continuously improves to obtain final identification model；S7, it after chirping of birds signal to be measured is obtained final feature vector according to step S1, S2, S4 processing, is entered into the final identification model of S6 and identifies the type and quantity of birds.
Description
Technical field
The present invention relates to birds identification technology field, more particularly to a kind of birds based on multiple features fusion transfer learning
Individual discrimination method.
Background technology
The activity of the mankind and the variation of the Nature have birdife environment direct or indirect influence, so as to bird
Class diversity constitutes a threat to.Birds quantity is the indispensable ecological variable of research Community Populations dynamic, selects it as life
Border suitability quantitative assessment can provide basis to formulate rational wet land protection and management planning.Thus birds identification for
The protection of birds and birdife environment is significant.
For example, the big reed warbler preference in east is built a nest in the reed clump to grow fine, it is within breeding period to nest domain environment
It is required that it is particularly stringent, it is very sensitive to the variation of external environment, because but rather indicate that the preferable instruction of Yeyahu Wetland environmental change
Species.The investigation generally use manual research method of the big reed warbler in east is largely dependent upon the vision and hearing of observer
Ability, dense vegetation can cause observer to the identification of distant place birds and count difficulty, reduce so as to cause the field of investigation, and
And result has certain uncertainty.Wetland is relatively muddy simultaneously, and pad it is difficult, and in most cases can not be by
The vehicles, require a great deal of time and energy.In recent years, infrared camera technology wild animal survery and monitoring at home
In be widely applied, this method can not monitor the birds individual being blocked, therefore be not suitable for the investigation of the big reed warbler in east.
In conclusion traditional investigation and monitoring method there are it is different the drawbacks of, it is necessary to study new method make up more than it is insufficient.Greatly
Reed warbler sings the start-up portion of sentence in a more conservative in vivo, the embodiment personal feature of Relatively centralized, as theoretical foundation,
Using sound collection equipment and individual identification software, based on the big reed warbler individual in song identification east, the big reed warbler in east is realized
The method of investigation and monitoring, can not only overcome disadvantages mentioned above, but also be high efficiency, on a large scale non-damaging, low interference, monitoring
Method has huge application prospect.
Currently used chirm sorting technique includes：1st, the sorting technique based on template matches, most it is representative just
It is dynamic time warping algorithm, although this method accuracy of identification is higher, operand is too big, influences recognition efficiency.2nd, it establishes
The disaggregated model of feature based realizes classification, and common model or method have hidden Markov model, gauss hybrid models, support
Vector machine, random forest, autonomic nervous network, k arest neighbors and integrated study etc., manual extraction is suitably poor in such method
Different feature is still a big bottleneck；Depth convolutional neural networks are introduced bird by the 3rd, sorting technique based on deep learning, such method
In song identification, the bottleneck of manual extraction chirm difference characteristic is broken through using its excellent learning ability.H.V.Koops is utilized
Deep neural network (Deep Neural Network, DNN) realizes the birds identification based on song, and compared different defeated
Fashionable recognition effect, the results showed that using mel-frequency cepstrum coefficient (the Mel Frequency Cepstrum of audio signal
Coefficient, MFCC), delta-MFCC and delta-delta-MFCC be combined input when, recognition effect is best.
I.Potamitis using sonagram feature the study found that be based on depth convolutional neural networks (Deep Convolutional
Neural Network, DCNN) when being classified, syllable characteristic is selected to sing the classifying quality of feature more than selection as input
It is good.K.J.Piczak compared the identification of the DCNN of 3 kinds of different structures using the Meier frequency domain power spectrogram of audio signal as input
Effect, the results showed that the size of input power spectrogram, the number of plies of network and network structure can all have an impact recognition effect；
By in same audio signal, averaged using different zones as recognition result when inputting, the accuracy rate of identification can be improved.
Using depth convolutional neural networks, the method based on song identification birds species can obtain spy by the way that training is automatic
Sign, but training depth convolutional neural networks need a large amount of sample data because birds species song there are areal variation not
Directly it can be trained and be verified with the song downloaded on the net, and the song sample acquired on the spot is limited, small sample training is deep
The neural convolutional network of degree is easy to cause over-fitting so that the accuracy of identification of model declines.In addition, it is based on gray-scale map in the prior art
Gray level co-occurrence matrixes ask for representing the characteristic value of birds difference, the feature of selection is single, can not have better universal performance
Invention content
The purpose of the present invention is to provide a kind of birds recognition methods based on multiple features fusion transfer learning, existing to solve
The problem of research object is single, computational efficiency is poor, high to sample requirement on the spot when identifying birds with the presence of technology.
In order to achieve the above objectives, the present invention uses following technical proposals：
A kind of birds individual discrimination method based on multiple features fusion transfer learning, including：
S1, known chirping of birds signal is pre-processed, including
Using loss of the preemphasis compensation chirping of birds signal on high-frequency energy, Wiener filtering wiping out background noise is carried out, it is right
Chirping of birds signal is split processing, removes mute area；
S2, framing and windowing process are carried out to pretreated chirping of birds signal, the chirping of birds signal of framing length is obtained, to obtaining
The chirping of birds signal of framing length carry out chirplet transformation, spread out into the lines of chirplet basic functions a series of
Property combination, using wavelet coefficient generate sonagram,
The linear frequency modulation morther wavelet basic function selected for：
Wherein, t is time, tcFor time centre, fcFor center frequency, △ t be the duration, c is linear frequency modulation rate；
S3, to depth convolutional neural networks Inception-ResNet-v2 processing, obtain pre-training model；
S4, the sonagram generated in S2 is input to the feature vector that different layers are obtained in the pre-training model of S3, to obtaining
The feature vectors of different layers merged, obtain final feature vector；
S5, final feature vector is input in support vector machines, be trained, obtain identification model；
S6, performance detection is carried out to obtained identification model, chooses final identification model；
S7, chirping of birds signal to be measured is obtained into final feature vector according to step S1, S2, S4 processing, by feature to
Amount is input in the final identification model of S6 and is identified, and exports the type and quantity of birds.
Preferably, the step S3 to depth convolutional neural networks handle and be included：Remove the depth convolutional Neural
The output layer of network；It will freeze for moving to the network of the characteristic extraction part of identification model.
Preferably, the step S4 chooses the feature of three different layers, obtains three differences to three feature pools respectively
Feature vector.
Preferably, the step S6 includes：By the use of the marked good song library of project team as training or identification data,
The accuracy of identification model is verified using ten folding cross-validation methods, according to accuracy rate (Precision), recall rate (Recall)
With the performance of three parametric synthesis evaluation identification models of F values.
Beneficial effects of the present invention are as follows：
Technical solution of the present invention considers the chirm Finite Samples acquired on the spot, utilizes deep learning model extraction
Multiple features merged, realize the birds individual discrimination method based on transfer learning, this method is small to sample requirement, while can
To greatly shorten the training time, recognition accuracy is improved, there is versatility.By using technical solution of the present invention to birds
Identification and quantitative approach, to study and protecting Avian diversity and birdife environment provides reliable convenient condition.
Description of the drawings
The specific embodiment of the present invention is described in further detail below in conjunction with the accompanying drawings.
Fig. 1 shows a kind of flow diagram of the birds individual discrimination method based on multiple features fusion transfer learning；
Fig. 2 shows the comparison diagrams using chirplet conversion sonagram.
Specific embodiment
In order to illustrate more clearly of the present invention, the present invention is done further with reference to preferred embodiments and drawings
It is bright.Similar component is indicated with identical reference numeral in attached drawing.It will be appreciated by those skilled in the art that institute is specific below
The content of description is illustrative and be not restrictive, and should not be limited the scope of the invention with this.
As shown in Figure 1, the invention discloses a kind of birds individual discrimination methods based on multiple features fusion transfer learning
Flow diagram.Known chirping of birds signal is pre-processed, framing, adding window, obtain framing length chirping of birds signal；Recycle line
Property Frequency Modulation Wavelet Transform, is spread out into a series of linear combination of chirplet basic functions, is generated using wavelet coefficient
Sonagram, wherein the linear frequency modulation morther wavelet basic function selected for：
In formula, t is the time, tcFor time centre, fcFor center frequency, △ t be the duration, c is linear frequency modulation rate；To depth
Convolutional neural networks Inception-ResNet-v2 processing removes the output layer of the depth convolutional neural networks, will use
Freezed to obtain pre-training model in the network for the characteristic extraction part for moving to identification model, sonagram is input to pre-training
The feature of different layers is chosen in model, obtains corresponding feature vector to different layers feature pool respectively, is obtained after fusion final
Feature vector；Final feature vector is input in support vector machines and is trained, obtains identification model, the identification to obtaining
Model carries out performance detection, by the use of the marked good song library of project team as the data of training or identification, is intersected using ten foldings
Proof method verifies the accuracy of identification model, according to three accuracy rate (Precision), recall rate (Recall) and F values ginsengs
Number synthesis evaluates the performance of identification model, chooses final identification model；By chirping of birds signal to be measured by above-mentioned steps handle, i.e., into
Row pretreatment, is input to neural network at framing adding window conversion sonagram, obtains the feature vector of different layers, is obtained after fusion final
Feature vector, feature vector is input in final identification model and distinguished, export the type and quantity of birds.
By the above method, the feature of the different layers of selection further can may be arranged combination fusion and be formed by the present invention
Different feature vectors analyzes corresponding recognition effect, feature the most suitable to be selected to combine, so as to obtain optimal identification mould
Type.
In order to further illustrate the present invention provides following examples.
To known east, the chirping of birds signal of big reed warbler pre-processes first, and the big reed warbler in east is compensated by preemphasis
Then loss of the chirping of birds signal on high-frequency energy carries out it Wiener filtering wiping out background noise, finally to chirping of birds signal into
Row dividing processing removes mute area.
According to the chirping of birds time span of the big reed warbler in east, it is 25ms that the present embodiment, which chooses frame length,.By pretreated chirping of birds
Signal carries out framing and windowing process, obtains the chirping of birds signal for the big reed warbler in east that multiple frame lengths are 25ms, chirplet
The sonagram being converted into has better recognition effect, as shown in Figure 2.The big reed warbler chirping of birds signal in east that obtained framing is grown is carried out
Chirplet converts CT, spreads out into a series of linear combination of chirplet basic functions, utilizes wavelet coefficient
Generate sonagram, the linear frequency modulation morther wavelet basic function selected for：
Wherein, t is time, tcFor time centre, fcFor center frequency, △ t be the duration, c is linear frequency modulation rate, thus
Go out multiple sonagrams.
The present embodiment chooses depth convolutional neural networks Inception-ResNet-v2, removes the depth convolutional Neural
The output layer of network will be freezed to obtain pre-training model for moving to the network of the characteristic extraction part of identification model,
The sonagram of the big reed warbler in above-mentioned east is input in pre-training model.The feature that depth convolutional neural networks different layers obtain is not
Together, the feature of more bottom is more local, more general, and more high-rise feature is then more global, more specific, in order to make full use of different layers
The feature of output chooses the feature of three different layers, three different feature vectors is obtained to three feature pools respectively, by it
Final feature vector is obtained after fusion.
Support vector machines (Support Vector Machine, SVM) can utilize small sample to realize efficiently classification, will most
Whole feature vector, which is input in support vector machines, to be trained, and identification model is obtained, to obtained identification model progressive
It can detect, by the use of the marked good song library of project team as the data of training or identification, be tested using ten folding cross-validation methods
The accuracy of identification model is demonstrate,proved, according to three accuracy rate (Precision), recall rate (Recall) and F values parametric synthesis evaluations
The performance of identification model chooses final identification model.
Chirping of birds signal to be measured is pre-processed, framing adding window conversion sonagram, is input to neural network, obtains different layers
Feature vector obtains final feature vector after fusion, chirping of birds signal characteristic vector to be measured is input to trained identification
It is identified in model, whether the type for judging bird is the big reed warbler in east, and then the big reed warbler in east is exported simultaneously if the big reed warbler in east
Quantity.
Technical solution of the present invention can solve the problems, such as the chirm Finite Samples acquired on the spot, utilize deep learning
The multiple features of model extraction are merged, and are realized the birds individual discrimination method based on transfer learning, are greatly shortened the training time,
Recognition accuracy is improved, while suitable for the identification of other birds, there is versatility.
Obviously, the above embodiment of the present invention be only to clearly illustrate example of the present invention, and not be pair
The restriction of embodiments of the present invention for those of ordinary skill in the art, may be used also on the basis of the above description
To make other variations or changes in different ways, all embodiments can not be exhaustive here, it is every to belong to this hair
The obvious changes or variations that bright technical solution is extended out are still in the row of protection scope of the present invention.
Claims (4)
1. a kind of birds individual discrimination method based on multiple features fusion transfer learning, which is characterized in that method and step includes：
S1, known chirping of birds signal is pre-processed, including
Using loss of the preemphasis compensation chirping of birds signal on high-frequency energy, Wiener filtering wiping out background noise is carried out, to chirping of birds
Signal is split processing, removes mute area；
S2, framing and windowing process are carried out to pretreated chirping of birds signal, obtains the chirping of birds signal of framing length, determine what is obtained
The chirping of birds signal of frame length carries out chirplet transformation, spreads out into linear group of chirplet basic functions a series of
It closes, sonagram is generated using wavelet coefficient,
The linear frequency modulation morther wavelet basic function selected for：
Wherein, t is time, tcFor time centre, fcFor center frequency, △ t be the duration, c is linear frequency modulation rate；
S3, to depth convolutional neural networks Inception-ResNet-v2 processing, obtain pre-training model；
S4, the sonagram generated in S2 is input to the feature vector that different layers are obtained in the pre-training model of S3, to acquisition not
The feature vector of same layer is merged, and obtains final feature vector；
S5, final feature vector is input in support vector machines, be trained, obtain identification model；
S6, performance detection is carried out to obtained identification model, chooses final identification model；
S7, chirping of birds signal to be measured is obtained into final feature vector according to step S1, S2, S4 processing, feature vector is defeated
Enter into the final identification model of S6 and be identified, export the type and quantity of birds.
2. a kind of birds individual discrimination method based on multiple features fusion transfer learning according to claim 1, feature
It is, the step S3 carries out depth convolutional neural networks processing and includes：
Remove the output layer of the depth convolutional neural networks；
It will freeze for moving to the network of the characteristic extraction part of identification model.
3. a kind of birds individual discrimination method based on multiple features fusion transfer learning according to claim 1, feature
It is, the step S4 chooses the feature of three different layers, obtains three different feature vectors to three feature pools respectively,
Fusion obtains final feature vector.
4. a kind of birds individual discrimination method based on multiple features fusion transfer learning according to claim 1, feature
It is, the step S6 includes
By the use of the marked good song library of project team as the data of training or identification, known using ten folding cross-validation methods to verify
The accuracy of other model, according to accuracy rate, the performance of three parametric synthesis evaluation identification models of recall rate and F values.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201810058223.1A CN108197591A (en) | 2018-01-22 | 2018-01-22 | A kind of birds individual discrimination method based on multiple features fusion transfer learning |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201810058223.1A CN108197591A (en) | 2018-01-22 | 2018-01-22 | A kind of birds individual discrimination method based on multiple features fusion transfer learning |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN108197591A true CN108197591A (en) | 2018-06-22 |
Family
ID=62590070
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201810058223.1A Pending CN108197591A (en) | 2018-01-22 | 2018-01-22 | A kind of birds individual discrimination method based on multiple features fusion transfer learning |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN108197591A (en) |
Cited By (14)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN109117732A (en) * | 2018-07-16 | 2019-01-01 | 国网江西省电力有限公司电力科学研究院 | A kind of transmission line of electricity relates to the identification of bird failure bird kind figure sound and control method |
| CN109145723A (en) * | 2018-07-09 | 2019-01-04 | 长江大学 | A kind of seal recognition methods, system, terminal installation and storage medium |
| CN109165636A (en) * | 2018-09-28 | 2019-01-08 | 南京邮电大学 | A kind of sparse recognition methods of Rare Birds based on component-level multiple features fusion |
| CN109979441A (en) * | 2019-04-03 | 2019-07-05 | 中国计量大学 | A kind of birds recognition methods based on deep learning |
| CN110164452A (en) * | 2018-10-10 | 2019-08-23 | 腾讯科技（深圳）有限公司 | A kind of method of Application on Voiceprint Recognition, the method for model training and server |
| CN110246504A (en) * | 2019-05-20 | 2019-09-17 | 平安科技（深圳）有限公司 | Birds sound identification method, device, computer equipment and storage medium |
| CN110399796A (en) * | 2019-09-02 | 2019-11-01 | 国网上海市电力公司 | A kind of electrical energy power quality disturbance recognition methods based on improvement deep learning algorithm |
| CN111048101A (en) * | 2020-01-15 | 2020-04-21 | 合肥慧图软件有限公司 | Biodiversity species analysis method based on voice recognition technology |
| CN111626093A (en) * | 2020-03-27 | 2020-09-04 | 国网江西省电力有限公司电力科学研究院 | Electric transmission line related bird species identification method based on sound power spectral density |
| CN113707159A (en) * | 2021-08-02 | 2021-11-26 | 南昌大学 | Electric network bird-involved fault bird species identification method based on Mel language graph and deep learning |
| CN114067368A (en) * | 2022-01-17 | 2022-02-18 | 国网江西省电力有限公司电力科学研究院 | Power grid harmful bird species classification and identification method based on deep convolution characteristics |
| CN116310894A (en) * | 2023-02-22 | 2023-06-23 | 中交第二公路勘察设计研究院有限公司 | Unmanned aerial vehicle remote sensing-based intelligent recognition method for small-sample and small-target Tibetan antelope |
| CN116722992A (en) * | 2023-02-22 | 2023-09-08 | 浙江警察学院 | Fraud website identification method and device based on multi-mode fusion |
| CN117727308A (en) * | 2024-02-18 | 2024-03-19 | 百鸟数据科技(北京)有限责任公司 | Mixed bird song recognition method based on deep migration learning |
Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20130132311A1 (en) * | 2011-11-18 | 2013-05-23 | Honeywell International Inc. | Score fusion and training data recycling for video classification |
| CN105631469A (en) * | 2015-12-18 | 2016-06-01 | 华南理工大学 | Bird image recognition method by multilayer sparse coding features |
| CN106682233A (en) * | 2017-01-16 | 2017-05-17 | 华侨大学 | Method for Hash image retrieval based on deep learning and local feature fusion |
| CN107066520A (en) * | 2017-02-14 | 2017-08-18 | 北京飞搜科技有限公司 | A kind of picture retrieval method and system based on multiple semantic hierarchies |
| CN107393542A (en) * | 2017-06-28 | 2017-11-24 | 北京林业大学 | A kind of birds species identification method based on binary channels neutral net |
-
2018
- 2018-01-22 CN CN201810058223.1A patent/CN108197591A/en active Pending
Patent Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20130132311A1 (en) * | 2011-11-18 | 2013-05-23 | Honeywell International Inc. | Score fusion and training data recycling for video classification |
| CN105631469A (en) * | 2015-12-18 | 2016-06-01 | 华南理工大学 | Bird image recognition method by multilayer sparse coding features |
| CN106682233A (en) * | 2017-01-16 | 2017-05-17 | 华侨大学 | Method for Hash image retrieval based on deep learning and local feature fusion |
| CN107066520A (en) * | 2017-02-14 | 2017-08-18 | 北京飞搜科技有限公司 | A kind of picture retrieval method and system based on multiple semantic hierarchies |
| CN107393542A (en) * | 2017-06-28 | 2017-11-24 | 北京林业大学 | A kind of birds species identification method based on binary channels neutral net |
Non-Patent Citations (1)
| Title |
|---|
| 张伟等: ""深度卷积神经网络特征提取用于地表覆盖分类初探"", 《中国图象图形学报》 * |
Cited By (18)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN109145723A (en) * | 2018-07-09 | 2019-01-04 | 长江大学 | A kind of seal recognition methods, system, terminal installation and storage medium |
| CN109117732A (en) * | 2018-07-16 | 2019-01-01 | 国网江西省电力有限公司电力科学研究院 | A kind of transmission line of electricity relates to the identification of bird failure bird kind figure sound and control method |
| CN109165636A (en) * | 2018-09-28 | 2019-01-08 | 南京邮电大学 | A kind of sparse recognition methods of Rare Birds based on component-level multiple features fusion |
| CN110164452A (en) * | 2018-10-10 | 2019-08-23 | 腾讯科技（深圳）有限公司 | A kind of method of Application on Voiceprint Recognition, the method for model training and server |
| CN109979441A (en) * | 2019-04-03 | 2019-07-05 | 中国计量大学 | A kind of birds recognition methods based on deep learning |
| CN110246504A (en) * | 2019-05-20 | 2019-09-17 | 平安科技（深圳）有限公司 | Birds sound identification method, device, computer equipment and storage medium |
| CN110399796A (en) * | 2019-09-02 | 2019-11-01 | 国网上海市电力公司 | A kind of electrical energy power quality disturbance recognition methods based on improvement deep learning algorithm |
| CN111048101A (en) * | 2020-01-15 | 2020-04-21 | 合肥慧图软件有限公司 | Biodiversity species analysis method based on voice recognition technology |
| CN111626093A (en) * | 2020-03-27 | 2020-09-04 | 国网江西省电力有限公司电力科学研究院 | Electric transmission line related bird species identification method based on sound power spectral density |
| CN111626093B (en) * | 2020-03-27 | 2023-12-26 | 国网江西省电力有限公司电力科学研究院 | Method for identifying related bird species of power transmission line based on sound power spectral density |
| CN113707159A (en) * | 2021-08-02 | 2021-11-26 | 南昌大学 | Electric network bird-involved fault bird species identification method based on Mel language graph and deep learning |
| CN113707159B (en) * | 2021-08-02 | 2024-05-03 | 南昌大学 | Power grid bird-involved fault bird species identification method based on Mel language graph and deep learning |
| CN114067368A (en) * | 2022-01-17 | 2022-02-18 | 国网江西省电力有限公司电力科学研究院 | Power grid harmful bird species classification and identification method based on deep convolution characteristics |
| CN116310894A (en) * | 2023-02-22 | 2023-06-23 | 中交第二公路勘察设计研究院有限公司 | Unmanned aerial vehicle remote sensing-based intelligent recognition method for small-sample and small-target Tibetan antelope |
| CN116722992A (en) * | 2023-02-22 | 2023-09-08 | 浙江警察学院 | Fraud website identification method and device based on multi-mode fusion |
| CN116310894B (en) * | 2023-02-22 | 2024-04-16 | 中交第二公路勘察设计研究院有限公司 | Unmanned aerial vehicle remote sensing-based intelligent recognition method for small-sample and small-target Tibetan antelope |
| CN117727308A (en) * | 2024-02-18 | 2024-03-19 | 百鸟数据科技(北京)有限责任公司 | Mixed bird song recognition method based on deep migration learning |
| CN117727308B (en) * | 2024-02-18 | 2024-04-26 | 百鸟数据科技(北京)有限责任公司 | Mixed bird song recognition method based on deep migration learning |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN108197591A (en) | A kind of birds individual discrimination method based on multiple features fusion transfer learning | |
| CN112163465B (en) | Fine-grained image classification method, fine-grained image classification system, computer equipment and storage medium | |
| CN112163461B (en) | Underwater target identification method based on multi-mode fusion | |
| CN107610707A (en) | A kind of method for recognizing sound-groove and device | |
| CN103870845B (en) | Novel K value optimization method in point cloud clustering denoising process | |
| CN110609320B (en) | Pre-stack seismic reflection pattern recognition method based on multi-scale feature fusion | |
| CN105488466B (en) | A kind of deep-neural-network and Acoustic Object vocal print feature extracting method | |
| Miao et al. | Underwater acoustic signal classification based on sparse time–frequency representation and deep learning | |
| CN110807760B (en) | Tobacco leaf grading method and system | |
| CN112613536B (en) | Near infrared spectrum diesel fuel brand recognition method based on SMOTE and deep learning | |
| CN106529667A (en) | Logging facies identification and analysis method based on fuzzy depth learning in big data environment | |
| CN110853656B (en) | Audio tampering identification method based on improved neural network | |
| CN114564982A (en) | Automatic identification method for radar signal modulation type | |
| WO2008016109A1 (en) | Learning data set optimization method for signal identification device and signal identification device capable of optimizing the learning data set | |
| CN112985574B (en) | High-precision classification identification method for optical fiber distributed acoustic sensing signals based on model fusion | |
| CN113780242A (en) | Cross-scene underwater sound target classification method based on model transfer learning | |
| CN108694953A (en) | A kind of chirping of birds automatic identifying method based on Mel sub-band parameter features | |
| CN105510970A (en) | Method for obtaining seismic facies optimal classification number | |
| CN104463194A (en) | Driver-vehicle classification method and device | |
| CN115545086B (en) | Migratable feature automatic selection acoustic diagnosis method and system | |
| CN111339935A (en) | Optical remote sensing picture classification method based on interpretable CNN image classification model | |
| CN101533475A (en) | Method for extracting feature of shape-adaptive neighborhood based remote sensing image | |
| CN114980122A (en) | Small sample radio frequency fingerprint intelligent identification system and method | |
| CN114782740A (en) | Remote sensing water quality monitoring method combining genetic optimization and extreme gradient promotion | |
| Wang et al. | A hierarchical birdsong feature extraction architecture combining static and dynamic modeling |
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 | ||
| RJ01 | Rejection of invention patent application after publication |
Application publication date: 20180622 |
|
| RJ01 | Rejection of invention patent application after publication |