CN106844614A - A kind of floor plan functional area system for rapidly identifying - Google Patents
A kind of floor plan functional area system for rapidly identifying Download PDFInfo
- Publication number
- CN106844614A CN106844614A CN201710036469.4A CN201710036469A CN106844614A CN 106844614 A CN106844614 A CN 106844614A CN 201710036469 A CN201710036469 A CN 201710036469A CN 106844614 A CN106844614 A CN 106844614A
- Authority
- CN
- China
- Prior art keywords
- floor plan
- image
- functional area
- functional
- system module
- 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
- G06F16/00—Information retrieval; Database structures therefor; File system structures therefor
- G06F16/50—Information retrieval; Database structures therefor; File system structures therefor of still image data
- G06F16/58—Retrieval characterised by using metadata, e.g. metadata not derived from the content or metadata generated manually
- G06F16/583—Retrieval characterised by using metadata, e.g. metadata not derived from the content or metadata generated manually using metadata automatically derived from the content
- G06F16/5838—Retrieval characterised by using metadata, e.g. metadata not derived from the content or metadata generated manually using metadata automatically derived from the content using colour
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/22—Matching criteria, e.g. proximity measures
-
- 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
Landscapes
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Data Mining & Analysis (AREA)
- Physics & Mathematics (AREA)
- General Engineering & Computer Science (AREA)
- General Physics & Mathematics (AREA)
- Life Sciences & Earth Sciences (AREA)
- Artificial Intelligence (AREA)
- Library & Information Science (AREA)
- Evolutionary Computation (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Biophysics (AREA)
- Bioinformatics & Computational Biology (AREA)
- Bioinformatics & Cheminformatics (AREA)
- Databases & Information Systems (AREA)
- Health & Medical Sciences (AREA)
- Biomedical Technology (AREA)
- Evolutionary Biology (AREA)
- Computational Linguistics (AREA)
- General Health & Medical Sciences (AREA)
- Molecular Biology (AREA)
- Computing Systems (AREA)
- Mathematical Physics (AREA)
- Software Systems (AREA)
- Image Analysis (AREA)
Abstract
The invention discloses a kind of floor plan functional area system for rapidly identifying, including image processing system module 100, image identification and detection system module 200 and floor plan cloud service system module 300, wherein, image identification and detection system module is connected with image processing system module, floor plan cloud service system module, is connected with image identification and detection system module.A kind of floor plan functional area system for rapidly identifying disclosed by the invention, it can reliably, rapidly recognize the functional area image information having in floor plan, meet the demand that people are retrieved to the function information of floor plan, and then improve work and the quality of the life of people, be conducive to widely popularization and application, be of great practical significance.
Description
Technical field
The present invention relates to technologies such as image processing field, field of image recognition, floor plan identification field, database cloud computings
Field, more particularly to a kind of floor plan functional area system for rapidly identifying.
Background technology
At present, continuing to develop with human sciences's technology, deep learning (Deep Learning) turns into engineering
One emerging field in habit field.In recent years, the application about deep learning is more and more wider, has been directed to speech recognition, figure
As fields such as identification, natural language processings.Deep learning simultaneously will continue to have influence on other keys of machine learning and artificial intelligence
Field.
Nowadays in artificial intelligence and the two fields of big data cloud computing, the appearance of deep learning first brings numerous
The change in field, the problem that conventional many cann't be solved is for example unmanned all to have become reality, image function region recognition
Detection field is no exception, and deep learning is strided forward to image recognition detection field, and big data cloud computing is similarly in addition
The possibility of various realizations is provided for other each fields.
Turn into the main carriers in internet along with picture, problem occurs therewith, when information is by literature record, mesh
Before, people can be easily found required content and arbitrarily edited by keyword search, and it is by scheming to work as information
When piece is recorded, such as when the information of house is represented by floor plan, existing technology is difficult to the function in floor plan
Information is retrieved, that is, be difficult to carry out the function information of each functional area such as bedroom, parlor, toilet, kitchen in floor plan
Retrieval, so as to have impact on the efficiency that user etc. finds key content from pictorial information, although the picture strip as floor plan
Come efficiently information record and mode of sharing, but but reducing people carries out the efficiency of information retrieval, it is impossible to meet people
The demand retrieved to the information of each functional area in floor plan.
Therefore, at present in the urgent need to developing a kind of technology, it can reliably, rapidly recognize the work(having in floor plan
Energy area image, meets the demand that people are retrieved to the function information of floor plan, and then improve the work and life of people
Quality.
The content of the invention
In view of this, it is an object of the invention to provide a kind of floor plan functional area system for rapidly identifying, its can reliably,
The functional area image information rapidly having in identification floor plan, meets what people were retrieved to the function information of floor plan
Demand, and then work and the quality of the life of people are improved, be conducive to widely popularization and application, anticipated with great production practices
Justice.
Therefore, the invention provides a kind of floor plan functional area system for rapidly identifying, including：
Image processing system module, for receiving the floor plan figure recognized the need for external image collecting device is gathered
Picture, then carries out definition judgment operation to the floor plan image, when its definition does not meet pre-conditioned, performs pretreatment
Operation, and image identification and detection system module will be sent to by the floor plan image of pretreatment operation, and it is clear to work as its
When degree meets pre-conditioned, image identification and detection system module is directly sent it to；
Image identification and detection system module, is connected with image processing system module, for receiving described image processing system
The floor plan image that system module is sent, and extract and identify the family that functional category is respectively marked with each functional area
Type figure image, is then sent to floor plan cloud service system module；
Floor plan cloud service system module, is connected with image identification and detection system module, for storing default house type
Chart database, and it is respectively marked with functional category on each functional area that described image recognition detection system module is sent is received
The floor plan image, on each the functional area image having in the floor plan image mark functional category difference
Marked in advance with each the functional area image having in the multiple floor plan images prestored in the house type chart database
The functional category of note carries out similarity comparison matching, when matching filters out the function class marked in advance on each functional area image
When other similarity is all higher than one or more floor plan image of preset value, judge to have in the floor plan image is every
The functional category mark marked on individual functional area image is accurate, then the institute of functional category will be respectively marked with each functional area
State floor plan image and be sent to user.
Wherein, the house type chart database is stored in advance in cloud server, and the house type chart database includes multiple
Each the functional area image having in floor plan image and each floor plan image and the corresponding relation between them, and
Mark has in advance on each described functional area image.
Wherein, in described image processing system modules, the pretreatment operation is operated for illumination compensation.
Wherein, described image recognition detection system module includes neural network submodule, neural metwork training submodule
Block and floor plan image detection identification submodule, wherein：
Neural network submodule, for setting up predetermined depth convolutional neural networks, the predetermined depth convolutional Neural
Network includes input layer, the ground floor that the floor plan image sent to described image processing system modules successively is processed
Hidden layer, second layer hidden layer, third layer hidden layer, the 4th layer of hidden layer and output layer；
Neural metwork training submodule, is connected with neural network submodule, for the multiple pre- biddings of collection in advance
Accurate floor plan image is input in the predetermined depth convolutional neural networks, and the predetermined depth convolutional neural networks are carried out
Training, the model convergence until causing the predetermined depth convolutional neural networks, completes the predetermined depth convolutional neural networks
Training；
Image detection recognizes submodule, is connected with image processing system module and neural metwork training submodule respectively,
For the floor plan image for sending described image processing system modules, the neural metwork training submodule is input to complete
Into in the predetermined depth convolutional neural networks of training, identification obtains each functional area having in the floor plan image
Image and the corresponding functional category data characteristics of each functional area image, while by the multiple functional category data characteristics
Be input in the default grader of the output layer carries out functional category classification respectively, then according to functional category classification results,
Mark function classification on each the functional area image having in the floor plan image, so as to obtain on each functional area
It is respectively marked with the floor plan image of functional category.
Wherein, the floor plan cloud service system module includes that floor plan database purchase submodule and functional area are contrasted
Matching output sub-module, wherein：
Floor plan database purchase submodule, for prestoring house type chart database, the house type chart database includes
Each the functional area image having in multiple floor plan images and each floor plan image and the corresponding relation between them,
And mark has in advance on each described functional area image；
Functional area contrast matching output sub-module, deposits with image identification and detection system module and house type chart database respectively
Storage submodule is connected, for receiving have in the floor plan image that described image recognition detection system module is sent complete
Portion's preset function regional image information, and by its with the house type chart database in each floor plan image for prestoring have
Whole preset function regional image informations carry out contrast matching, according to contrast matching result, filter out the corresponding house type
One or more floor plan image in chart database, is then sent to user.
Wherein, the floor plan cloud service system module also includes that floor plan updates submodule, and the floor plan updates submodule
Block is connected with functional area contrast matching output sub-module, for being obtained from functional area contrast matching output sub-module
The floor plan image that functional category is respectively marked with each functional area is taken, and is sent to floor plan database purchase
Module is stored.
Wherein, the floor plan cloud service system module is cloud server.
The technical scheme that the present invention is provided more than, compared with prior art, the invention provides a kind of house type
Figure functional area system for rapidly identifying, it can reliably, rapidly recognize the functional area image information that floor plan has, and meet
The demand that people are retrieved to the function information of floor plan, and then work and the quality of the life of people are improved, be conducive to extensively
Ground popularization and application, are of great practical significance.
Brief description of the drawings
A kind of block diagram of floor plan functional area system for rapidly identifying that Fig. 1 is provided for the present invention
Fig. 2 is a schematic diagram for floor plan image to be detected in a particular embodiment；
Fig. 3 is in a particular embodiment, a kind of floor plan functional area system for rapidly identifying provided by the present invention enters
Schematic diagram after row identification.
Specific embodiment
In order that those skilled in the art more fully understand the present invention program, below in conjunction with the accompanying drawings with implementation method to this
Invention is described in further detail.
A kind of block diagram of floor plan functional area system for rapidly identifying that Fig. 1 is provided for the present invention.
Referring to a kind of floor plan functional area system for rapidly identifying that Fig. 1, the present invention are provided, including image processing system mould
Block 100, image identification and detection system module 200 and floor plan cloud service system module 300, wherein：
Image processing system module 100, is adopted for receiving external image collecting device (such as mobile phone or computer)
The floor plan image recognized the need for collection, then carries out definition judgment operation, when its definition is not inconsistent to the floor plan image
When closing pre-conditioned, pretreatment operation is performed, and image recognition will be sent to by the floor plan image of pretreatment operation
Detecting system module 200, and when its definition meets pre-conditioned, directly send it to image identification and detection system module
200；
Image identification and detection system module 200, its as deep learning system module, with image processing system module
100 are connected, and for receiving the floor plan image that described image processing system modules 100 are sent, and extract and identify
The floor plan image of functional category is respectively marked with each functional area, and (functional area can include bedroom, parlor, health
Between, kitchen, the various functions region such as window, the functional category can include bedroom function, parlor function, toilet function, kitchen
The various functions such as room function, window function classification), it is then sent to floor plan cloud service system module 300；
Floor plan cloud service system module 300, is connected with image identification and detection system module 200, default for storing
House type chart database, and marked on each functional area that described image recognition detection system module 200 is sent is received
The floor plan image of functional classification, the work(marked on each the functional area image having in the floor plan image
Can classification respectively with each functional area figure for having in multiple floor plan images for prestoring in the house type chart database
As the functional category of upper advance mark carries out similarity comparison matching, marked in advance on each functional area image when matching is filtered out
When the similarity of the functional category of note is all higher than one or more floor plan image of preset value (for example, 98%), institute is judged
Stating the functional category mark marked on each the functional area image having in floor plan image, accurate (this is the knot of identification
Really), the floor plan image that functional category then will be respectively marked with each functional area is sent to user and (for example directly transmits
The mobile terminals such as mobile phone, panel computer to user).
Wherein, the house type chart database is preferably and is stored in advance in cloud server, the house type chart database bag
Include each functional area image for having in multiple (preferably magnanimity) floor plan images and each floor plan image and they
Between corresponding relation (i.e. mapping relations, such as one-to-one relationship or one-to-many relation), and each described function
Mark has in advance on area image.
In the present invention, the external image collecting device can have IMAQ and transfer function for any one
The mobile terminals such as equipment, such as mobile phone and panel computer or computer PC.
In the present invention, in described image processing system modules 100, the pretreatment operation is preferably illumination compensation behaviour
Make, therefore, improved by illumination compensation and know the need for external image collecting device (such as mobile phone or computer) is gathered
The quality of other floor plan image so that the identification of floor plan image is improved, and with definition higher, finally again pre-
Image after treatment is sent to image identification and detection system module 200, is identified in image identification and detection system module 200
With feature extraction.
In the present invention, it is necessary to explanation, described image processing system modules 100 are mainly by granny rag Lars energy side
The existing Approach for detecting image sharpness such as method estimates definition (the granny rag Lars energy side of the floor plan image
Method estimates the definition of image especially by the energy gradient distribution of the Ei of picture).
Implement, it is described it is pre-conditioned can be according to any setting be carried out the need for user in advance.For example, when described
When image processing system module 100 judges the definition of floor plan image by granny rag Lars ENERGY METHOD, the default bar
Part can be the granny rag Lars energy threshold of default image, i.e., the granny rag Lars energy threshold according to default image is sentenced
It is disconnected, when the granny rag Lars energy value of the floor plan image that external image collecting device is gathered is higher than this threshold value, it is judged as
Picture rich in detail, is otherwise judged as that blurred picture then carries out pretreatment operation again.
In the present invention, for described image recognition detection system module 200, it include neural network submodule,
Neural metwork training submodule and floor plan image detection identification submodule, these three submodules carry out depth convolutional Neural respectively
The treatment of the detection identification process for setting up process, the training process of depth convolutional neural networks and floor plan image of each layer of network
Operation, wherein：
Neural network submodule, for setting up predetermined depth convolutional neural networks (Convolutional Neural
Network, CNN), the predetermined depth convolutional neural networks include described image processing system modules 100 are sent successively
Input layer that the floor plan image is processed, ground floor hidden layer, second layer hidden layer, third layer hidden layer, the 4th layer
Hidden layer and output layer；
Neural metwork training submodule, is connected with neural network submodule, for the multiple pre- biddings of collection in advance
Accurate floor plan image (the floor plan image of the size that such as user specifies) is input to the predetermined depth convolutional Neural
In network, the predetermined depth convolutional neural networks are trained, until causing the predetermined depth convolutional neural networks
Model is restrained, and completes the training of the predetermined depth convolutional neural networks；
Image detection recognizes submodule, is connected with image processing system module 100 and neural metwork training submodule respectively
Connect, for the floor plan image for sending described image processing system modules 100, be input to neural metwork training
Module is completed in the predetermined depth convolutional neural networks of training, and identification obtains each work(having in the floor plan image
Energy area image and the corresponding functional category data characteristics of each functional area image (being embodied in depth convolution feature),
The multiple functional category data characteristics is input to default grader (the specially softmax classification of the output layer simultaneously
Device) in carry out functional category classification respectively, then according to functional category classification results, what is had in the floor plan image is every
Mark function classification on individual functional area image, so as to obtain the floor plan that functional category is respectively marked with each functional area
Image.
In the present invention, it is necessary to explanation, the corresponding functional category data characteristics of each functional area image, refers to every
The convolution feature of the image that individual functional area image information learns in the predetermined depth convolutional neural networks, this feature
The feature of the responses such as the pixel characteristic and structural edge of each functional area image is contained, is abstract characteristics, for characterizing not
Difference in functionality classification corresponding to congenerous area image.
In the present invention, implement, the functional category classification can be specifically included：Bedroom function, parlor function,
Various functional category classification such as kitchen function, parlor function, dining room function, toilet function.
For the present invention, the predetermined depth convolution of this present system is designed by neural network submodule first
Neural network structure, is set in neural metwork training submodule using previous module (i.e. neural network submodule) afterwards
The predetermined depth convolutional neural networks model set carries out the training of floor plan image, and then extracts corresponding characteristics of image,
The model containing the accordingly corresponding functional category data characteristics of each functional area is obtained, finally using neural metwork training submodule
The model for obtaining, realizes the detection and identification to the floor plan image to be detected in image detection identification submodule.
In the present invention, it is necessary to explanation, first, neural network submodule is only intended to predetermined depth convolution god
It is by containing that the neural network submodule is demarcated in neural metwork training submodule through the structure of network
The data of each functional area of corresponding each function information (i.e. functional category data characteristics), for neural network
The model that module is set up is trained and feature extraction, is finally detected identification, completion training predetermined depth convolution
Neural network model, then using the predetermined depth convolutional neural networks model pair for training in image detection identification submodule
Floor plan image to be detected detected, obtains testing result image.
In the present invention, implement, it is necessary to illustrate, for neural network submodule, the default volume
Product neutral net be based on the current main-stream deep learning detector faster improved depth convolutional networks of rcnn, the present invention from
Need to train factor of both substantial amounts of house type diagram data and time efficiency to consider, make full use of current main-stream to detect network
(the region suggestion network RPN structures of specially faster rcnn), and devise input layer, middle four layers of convolution hidden layer and
Output layer.First, the present invention takes full advantage of RPN network structure features, and object boundary is predicted simultaneously in each position to image
With target loss score, and high-quality region Suggestion box detection is reached using training method end to end；In addition, this hair
Bright network input layer is used for node (data) information for the floor plan image for being input into multiple preset standards, ground floor hidden layer bag
It is that the corresponding functional category data characteristics of function area image (is embodied in depth in floor plan image containing 128 input values
Degree convolution feature, by manually presetting) node (or other default multiples), the second layer include input value be function
The nodes of the corresponding functional category data characteristics of area image are 64 (or other default multiples), and third layer includes defeated
It is 32 (or other default multiples), the 4th to enter the nodes that value is the corresponding functional category data characteristics of functional area image
Layer hidden layer include input value be the corresponding functional category data characteristics of functional area image nodes for 16 (or its
He is default multiple), it is the corresponding functional category number of function area image in floor plan image that output layer includes 16 output valves
According to the node of feature.
It should be noted that in predetermined depth convolutional neural networks of the invention, each node is using the artificial side for calculating
Formula sets corresponding Mathematical Modeling and relevant parameter, the input value of each node sets corresponding floor plan for needed in input layer
The corresponding functional category data characteristics of each functional area, four hidden layers also have the input value of corresponding node in last output layer
Respectively corresponding functional category data feature values of regional function image in the floor plan image of last layer output, every layer all in addition
Corresponding weighting parameter ω and offset parameter κ is set, input and output relation between each layer are expressed as：Y=ω x+ κ,
In formula, x represents input neuron, and y represents output neuron, and w is weight, and κ is biasing.
In addition, in the predetermined depth convolutional neural networks that the present invention sets up, it is possible to use softmax graders are exported
Layer carries out identification and the work(of the corresponding functional category data characteristics of each functional area image that last floor plan image has
Can classify, so as to realize the classification to multiple functional areas in floor plan image.
In the present invention, implement, the effect of grader is extracted according to above predetermined depth convolutional neural networks
Feature, the corresponding functional category data characteristics of each functional area image having to the floor plan image carries out function class
Do not classify.Implement, the present invention can use softmax graders.Each functional areas that the floor plan image has
The corresponding functional category data characteristics of area image can be according to pre-setting in the system of the present invention the need for user, the work(
Can classification can including parlor function, bedroom function, kitchen function, toilet function and functional category etc. balcony function, certainly,
According to the difference of specific house type, other classifications, such as storeroom functional category can also be increased.
For softmax graders, its probability distribution that can calculate different classes of depth convolution feature, according to difference
Probability distribution judges the classification of each functional area image in floor plan image.Specific operating process is the output of preceding layer
It one is characteristic value to be, by these characteristic values being multiplied by into different weights and then being normalized, you can obtain different work(
The probability distribution of energy area image.
In the present invention, implement, for neural metwork training submodule, it uses the error back propagation of optimization
BP algorithm carries out the training to predetermined depth convolutional neural networks, first before training, by given threshold and weights, makes threshold
Value and weights carry out the random initializtion in the range of from -1 to 1, and when data are fitted, the present invention is utilizing the double cosine of Sigmoid just
Function is cut as excitation function, drops it off after intermediate layer exports to ensure that the value of output node can be in (0,1) this scope
Interior, in addition, the present invention can be by setting a loss function loss come error in judgement, the computing formula of loss function is：
Wherein, Y0For the prediction of predetermined depth convolutional neural networks is exported, and YtrueOutput is demarcated for corresponding, when last
Demarcation output YtrueWith prediction output Y0When falling far short, at this moment corresponding loss function loss will be very big, predetermined depth volume
Product neutral net will carry out error-duration model to update the model parameter of network, when predetermined depth convolutional neural networks often train one
Secondary, the weighting parameter ω and offset parameter κ of corresponding each layer will update once, and then last demarcation is exported YtrueWith it is pre-
Survey output Y0Difference is less and less, when predetermined depth convolutional neural networks are by after repeatedly training, at this moment loss will be less than one
Determine threshold value, predetermined depth convolutional neural networks stop training, and training process now terminates, complete the predetermined depth convolution god
Through the training of network.
In the present invention, implement, submodule is recognized for image detection, it is based on neural metwork training submodule
The predetermined depth convolutional neural networks for training, to by described in the pretreatment of described image processing system modules 100
Floor plan image or the floor plan image for not pre-processing and directly sending are detected that identification obtains the floor plan figure
Each the functional area image having as in and the corresponding functional category data characteristics (specific manifestation of each functional area image
It is depth convolution feature), while the multiple functional category data characteristics to be input to the softmax graders of the output layer
In carried out functional category classification respectively, then according to functional category classification results, have in the floor plan image
Mark function classification on each functional area image, so as to obtain the house type that functional category is respectively marked with each functional area
Figure image, finally gives classification and the recognition result of the function information of each functional area image that each floor plan image has,
Can recognize the function information for obtaining each floor plan image (that is, according to the function of being marked on each functional area
Classification, it is possible to know the function situation of function situation that each floor plan includes and specific internal each functional area).
It should be noted that for the present invention, the mechanism of image identification and detection system module 200 includes neutral net
Each layer sets up process, neural network training process and image detection identification process, and the present invention can be by the mistake using optimization
Differ from backpropagation (BP) algorithm to accelerate the convergence rate of training process, and then avoid being absorbed in part because training substantial amounts of sample
Minimum situation.
In the present invention, for the floor plan cloud service system module 300, it includes floor plan database purchase submodule
Block and functional area contrast matching output sub-module, wherein：
Floor plan database purchase submodule, (preferably prestores beyond the clouds for prestoring house type chart database
In server), the house type chart database includes tool in multiple (preferably magnanimity) floor plan images and each floor plan image
Each functional area image for having and corresponding relation (i.e. mapping relations, such as one-to-one relationship or between them
To many relations), and mark has in advance on each described functional area image.
Functional area contrast matching output sub-module, respectively with image identification and detection system module 200 and house type diagram data
Library storage submodule is connected, for receiving the floor plan image that described image recognition detection system module 200 is sent in
The whole preset function regional image informations having, and by its with the house type chart database in each floor plan for prestoring
Whole preset function regional image informations that image has carry out contrast matching, according to contrast matching result, filter out corresponding
One or more floor plan image in the house type chart database (tie by the recognition result and retrieval obtained after as matching
Really), it is then sent to user's (being for example transmitted directly to the mobile terminals such as mobile phone, the panel computer of user).
It should be noted that for the present invention, implement, floor plan number can be set up in server beyond the clouds in advance
According to storehouse, the house type chart database includes all being preset in multiple (preferably magnanimity) floor plan images and each floor plan image
Functional area image information and corresponding relation (i.e. mapping relations, such as one-to-one relationship or one-to-many between them
Relation).Then, matching output sub-module is contrasted by functional area to recognize the required described image for contrast matching
The functional category marked on each the functional area image having in the floor plan image that detecting system module 200 is sent,
It is pre- with each functional area image for having in multiple floor plan images for prestoring in the house type chart database respectively
The functional category for first marking carries out similarity comparison matching, when the similarity that matching filters out each preset function region is all higher than
One or more floor plan image of default value (for example, 98%) (tie by the recognition result and retrieval obtained after as matching
When really), judge the functional category mark marked on each functional area image for having in the floor plan image it is accurate (this i.e.
It is the result of identification), the floor plan image that functional category then will be respectively marked with each functional area is sent to user's (example
Such as it is transmitted directly to mobile phone, the panel computer mobile terminal of user).
For the present invention, implement, in the floor plan functional area system for rapidly identifying that the present invention is provided, for institute
Floor plan cloud service system module 300 is stated, it also includes that floor plan updates submodule, and the floor plan updates submodule and functional areas
Domain contrast matching output sub-module is connected, for obtaining each function from functional area contrast matching output sub-module
The floor plan image of functional category is respectively marked with region, and is sent to the floor plan database purchase submodule and deposited
Storage (i.e. as new floor plan image, realizing data sync storage).
Therefore, in the floor plan functional area system for rapidly identifying that the present invention is provided, floor plan functional area can be allowed fast
Fast identifying system has the function of self-teaching again, can be the floor plan image for just carrying out image retrieval identification
Practise, and the new floor plan image for learning is added in the house type chart database stored in cloud server.
It should be noted that for the present invention, floor plan cloud service system module 300 can be to image identification and detection system
The repertoire area image having in the floor plan image that module 200 is transmitted is analyzed and compares, floor plan of the invention
The house type diagram data of magnanimity is covered in database such that it is able to effectively the floor plan image information for transmitting is screened in real time
With compare analysis, while have self-teaching, self-management and can receive from a large amount of mobile terminals request function, carry out and
When effectively to user collection input floor plan image carry out function information retrieval, in house type chart database recognize retrieval obtain
Obtain one or more house type matched with the functional category marked on each functional area in the floor plan image being input into
Figure image, therefore, the present invention can reliably, rapidly recognize the functional area image information having in floor plan, meet people
The demand retrieved to the function information of floor plan.
For a kind of floor plan functional area system for rapidly identifying that the present invention is provided, it is the entirety that user is serviced
Flow is as follows：
Firstly, it is necessary to the user for understanding each functional area function information of floor plan in detail can select according to the situation of oneself
Select at computer PC ends or gathered in mobile terminal and transmitted the floor plan functional areas that floor plan image is provided to the present invention
In the system for rapidly identifying of domain；
Then, the floor plan functional area system for rapidly identifying for being provided for the present invention, image processing system mould therein
Block 100 can be judged the readability of floor plan image first after the floor plan image that user is transmitted is received, if passing
The readability of floor plan image come meets pre-conditioned, then be transferred directly in image identification and detection system module 200 with
Carry out the extraction of functional area characteristics of image (i.e. functional area image information) and identification, and if the floor plan image for transmitting
Identification is poor, and readability does not meet pre-conditioned, then floor plan image first is carried out into pretreatment operation, mainly to figure
As carrying out illumination compensation operation, the quality of floor plan image is improved by illumination compensation, then by after pretreatment operation
Floor plan image be sent to image identification and detection system module 200；
Then, image identification and detection system module 200 is receiving the floor plan figure that image processing system module 100 is transmitted
As after information, being detected that this model is base using the optimized floor plan detection neural network model for training of the present invention
In the depth convolutional Neural of the high-quality rapid extraction characteristics of image of current main-stream deep learning detector faster rcnn designs
Network model；
Then, in image identification and detection system module 200, functional class is marked on each functional area detecting and exporting
After other floor plan image, the result of detection can be sent to floor plan cloud service system mould by image identification and detection system module 200
Block 300 carries out further screening and compares analysis；
Then, floor plan cloud service system module 300 is receiving the house type that image identification and detection system module 200 is transmitted
After figure functional area recognition result, various floor plan functional areas are carried out from the floor plan Database Systems in default high in the clouds first
The index in domain；
Then, after floor plan view data index, the data and image identification and detection system module 20 that then will be indexed
The result of each functional area of floor plan of recognition detection screened and compared, further to lift the identification of whole system
The result of floor plan functional area recognition detection, is finally fed back to user by accuracy rate.Meanwhile, floor plan cloud service system module
The image information of the new each functional area of floor plan for just having recognized, can be added to cloud by 300 functions with self-teaching
It is updated in client database and storage.
As shown in Figure 2 and Figure 3, the floor plan functional area based on deep learning for being provided according to the present invention quickly recognizes and is
System, the floor plan image that it is detected still can reach each functional area of floor plan under complicated and diversified function with Shandong nation
The purpose of identification is detected, accurate marker goes out the functional category of each functional area wherein, and Fig. 2 is floor plan image to be detected,
Fig. 3 is the testing result schematic diagram of floor plan functional area system for rapidly identifying of the present invention based on deep learning.
Therefore, understood based on above technical scheme, the present invention includes following beneficial effect：
First, the present invention solves prior art and is difficult to retrieve the function information of picture, and user is from floor plan
The low problem of functions area information efficiency is found in information；
Secondly, the module of deep learning is innovatively added, depth convolutional neural networks is made full use of to Efficient image rate
Feature extraction and recognition advantage, by projected depth convolution networking model, effectively analysis shows user needed for extracting
Each functional area of floor plan image feature information；
Again, the present invention is aided in using big data cloud computing technology simultaneously, to having extracted functional character information
Carry out further screening to compare to increase the accuracy rate of identification, and testing result fed back into user in real time；
Finally, the utilization scene of the floor plan functional area system for rapidly identifying based on deep learning that the present invention is provided is non-
Often extensively, the present invention can be not only the succinct floor plan functional area display platform of realtor's provides convenient, can make to sell
The attendant in room easily and efficiently shows information of selling house to house-purchase user, meanwhile, the present invention can also be at mobile terminal or PC ends
On effectively to house-purchase user push clearly floor plan functional area information show, user is visually known house-purchase
Information；Additionally, we are bright can to complete data interaction by communications service and floor plan cloud service system module；Have simultaneously certainly
I learn, self-management and can receive from a large amount of mobile terminals ask function, can timely and effectively by floor plan functional areas
The recognition result in domain feeds back to user.
In the present invention, implement, described image processing system modules 100 and image identification and detection system module
200 can be central processor CPU, digital signal processor DSP or single-chip microprocessor MCU, or be cloud server.
In the present invention, implement, the floor plan cloud service system module 300 can be cloud server, by
The data storage (such as hard disk) of cloud server prestores the house type chart database.
In sum, compared with prior art, the invention provides a kind of floor plan functional area system for rapidly identifying,
It can reliably, rapidly recognize the functional area image information having in floor plan, meet function letter of the people to floor plan
The demand that breath is retrieved, and then work and the quality of the life of people are improved, be conducive to widely popularization and application, with great
Production practices meaning.
By using the technology that the present invention is provided, can cause that people work and the convenience of life obtains very big carrying
Height, drastically increases the living standard of people.
The above is only the preferred embodiment of the present invention, it is noted that for the ordinary skill people of the art
For member, under the premise without departing from the principles of the invention, some improvements and modifications can also be made, these improvements and modifications also should
It is considered as protection scope of the present invention.
Claims (7)
1. a kind of floor plan functional area system for rapidly identifying, it is characterised in that including：
Image processing system module, for receiving the floor plan image recognized the need for external image collecting device is gathered, so
Definition judgment operation is carried out to the floor plan image afterwards, when its definition does not meet pre-conditioned, pretreatment operation is performed,
And image identification and detection system module will be sent to by the floor plan image of pretreatment operation, and work as its definition and meet
When pre-conditioned, image identification and detection system module is directly sent it to；
Image identification and detection system module, is connected with image processing system module, for receiving described image processing system mould
The floor plan image that block is sent, and extract and identify the floor plan that functional category is respectively marked with each functional area
Image, is then sent to floor plan cloud service system module；
Floor plan cloud service system module, is connected with image identification and detection system module, for storing default floor plan number
According to storehouse, and the institute of functional category is respectively marked with each functional area that described image recognition detection system module is sent is received
State floor plan image, on each the functional area image having in the floor plan image mark functional category respectively with institute
State what is marked in advance on each the functional area image having in the multiple floor plan images prestored in house type chart database
Functional category carries out similarity comparison matching, when matching filters out the functional category marked in advance on each functional area image
When similarity is all higher than one or more floor plan image of preset value, each work(having in the floor plan image is judged
The functional category mark marked on energy area image is accurate, then the family of functional category will be respectively marked with each functional area
Type figure image is sent to user.
2. floor plan functional area system for rapidly identifying as claimed in claim 1, it is characterised in that the house type chart database
It is stored in advance in cloud server, the house type chart database includes tool in multiple floor plan images and each floor plan image
Marked in advance on each the functional area image having and the corresponding relation between them, and each described functional area image
Functional classification.
3. floor plan functional area system for rapidly identifying as claimed in claim 1, it is characterised in that in described image processing system
In system module, the pretreatment operation is operated for illumination compensation.
4. floor plan functional area system for rapidly identifying as claimed in claim 1, it is characterised in that described image recognition detection
System module includes neural network submodule, neural metwork training submodule and floor plan image detection identification submodule,
Wherein：
Neural network submodule, for setting up predetermined depth convolutional neural networks, the predetermined depth convolutional neural networks
The input layer that is processed including the floor plan image sent to described image processing system modules successively, ground floor are hidden
Layer, second layer hidden layer, third layer hidden layer, the 4th layer of hidden layer and output layer；
Neural metwork training submodule, is connected with neural network submodule, for the multiple preset standards of collection in advance
Floor plan image is input in the predetermined depth convolutional neural networks, and the predetermined depth convolutional neural networks are instructed
Practice, the model convergence until causing the predetermined depth convolutional neural networks completes the predetermined depth convolutional neural networks
Training；
Image detection recognizes submodule, is connected with image processing system module and neural metwork training submodule respectively, is used for
The floor plan image that described image processing system modules are sent, is input to the neural metwork training submodule and completes instruction
In the experienced predetermined depth convolutional neural networks, identification obtains each the functional area image having in the floor plan image
And the corresponding functional category data characteristics of each functional area image, while the multiple functional category data characteristics is input into
Functional category classification is carried out respectively in the default grader of the output layer, then according to functional category classification results, in institute
Mark function classification on each the functional area image having in floor plan image is stated, so as to obtain be marked on each functional area
The floor plan image of functional classification.
5. the floor plan functional area system for rapidly identifying as any one of Claims 1-4, it is characterised in that described
Floor plan cloud service system module includes floor plan database purchase submodule and functional area contrast matching output sub-module, its
In：
Floor plan database purchase submodule, for prestoring house type chart database, the house type chart database includes multiple
Each the functional area image having in floor plan image and each floor plan image and the corresponding relation between them, and
Mark has in advance on each described functional area image；
Functional area contrast matching output sub-module, respectively with image identification and detection system module and floor plan database purchase
Module is connected, for receiving the whole having in the floor plan image that described image recognition detection system module is sent in advance
If functional area image information, and by its with the house type chart database in each floor plan image for prestoring have it is complete
Portion's preset function regional image information carries out contrast matching, according to contrast matching result, filters out the corresponding floor plan number
According to one or more floor plan image in storehouse, user is then sent to.
6. floor plan functional area system for rapidly identifying as claimed in claim 5, it is characterised in that the floor plan cloud service
System module also includes that floor plan updates submodule, and the floor plan updates submodule and matches output sub-module with functional area contrast
It is connected, be respectively marked with functional category on each functional area for is obtained from functional area contrast matching output sub-module
The floor plan image, and be sent to the floor plan database purchase submodule and stored.
7. floor plan functional area system for rapidly identifying as claimed in claim 1, it is characterised in that the floor plan cloud service
System module is cloud server.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201710036469.4A CN106844614A (en) | 2017-01-18 | 2017-01-18 | A kind of floor plan functional area system for rapidly identifying |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201710036469.4A CN106844614A (en) | 2017-01-18 | 2017-01-18 | A kind of floor plan functional area system for rapidly identifying |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN106844614A true CN106844614A (en) | 2017-06-13 |
Family
ID=59124445
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201710036469.4A Pending CN106844614A (en) | 2017-01-18 | 2017-01-18 | A kind of floor plan functional area system for rapidly identifying |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN106844614A (en) |
Cited By (26)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN107545306A (en) * | 2017-07-05 | 2018-01-05 | 安徽奇智科技有限公司 | A kind of big data analysis platform based on cloud computing |
| CN108009598A (en) * | 2017-12-27 | 2018-05-08 | 北京诸葛找房信息技术有限公司 | Floor plan recognition methods based on deep learning |
| CN108268748A (en) * | 2018-04-26 | 2018-07-10 | 南京信息工程大学 | A kind of residence model figure electrical design method based on machine learning |
| CN108764022A (en) * | 2018-04-04 | 2018-11-06 | 链家网（北京）科技有限公司 | A kind of image-recognizing method and system |
| CN108984765A (en) * | 2018-07-20 | 2018-12-11 | 我要家网络科技有限公司 | It is a kind of based on intelligent image analysis floor plan look for room method |
| CN109711443A (en) * | 2018-12-14 | 2019-05-03 | 平安城市建设科技(深圳)有限公司 | Floor plan recognition methods, device, equipment and storage medium neural network based |
| CN109740506A (en) * | 2018-12-29 | 2019-05-10 | 广东三维家信息科技有限公司 | A kind of house type image-recognizing method and device |
| CN109993797A (en) * | 2019-04-04 | 2019-07-09 | 广东三维家信息科技有限公司 | Door and window method for detecting position and device |
| CN110059721A (en) * | 2019-03-16 | 2019-07-26 | 平安城市建设科技(深圳)有限公司 | Floor plan area recognizing method, device, equipment and computer readable storage medium |
| CN110059690A (en) * | 2019-03-28 | 2019-07-26 | 广州智方信息科技有限公司 | Floor plan semanteme automatic analysis method and system based on depth convolutional neural networks |
| CN110059750A (en) * | 2019-04-17 | 2019-07-26 | 广东三维家信息科技有限公司 | House type shape recognition process, device and equipment |
| CN110111426A (en) * | 2019-04-18 | 2019-08-09 | 贝壳技术有限公司 | A kind of determination method and apparatus in sound separate pattern house |
| CN110197225A (en) * | 2019-05-28 | 2019-09-03 | 广东三维家信息科技有限公司 | House type spatial match method and system based on deep learning |
| CN110390304A (en) * | 2019-07-24 | 2019-10-29 | 广东南方数码科技股份有限公司 | Automatic classification method, device, electronic equipment and storage medium |
| CN110390153A (en) * | 2019-07-15 | 2019-10-29 | 贝壳技术有限公司 | Generation method, device and the equipment of layout structure modification scheme, storage medium |
| CN110443895A (en) * | 2019-06-27 | 2019-11-12 | 贝壳技术有限公司 | House illumination scheme generation method and device |
| CN110728235A (en) * | 2019-10-11 | 2020-01-24 | 广东三维家信息科技有限公司 | House type area marking method and device |
| CN110826627A (en) * | 2019-11-06 | 2020-02-21 | 广东三维家信息科技有限公司 | Image similarity measuring method and device and electronic equipment |
| CN111291649A (en) * | 2020-01-20 | 2020-06-16 | 广东三维家信息科技有限公司 | Image recognition method and device and electronic equipment |
| CN111400519A (en) * | 2019-12-05 | 2020-07-10 | 江苏艾佳家居用品有限公司 | Efficient customized massive similar house type searching method |
| CN111798340A (en) * | 2020-06-24 | 2020-10-20 | 北京字节跳动网络技术有限公司 | House type graph identification method and device and electronic equipment |
| CN112116613A (en) * | 2020-09-25 | 2020-12-22 | 贝壳技术有限公司 | Model training method, image segmentation method, image vectorization method and system thereof |
| CN114554503A (en) * | 2020-11-24 | 2022-05-27 | 中移物联网有限公司 | Method and device for confirming networking information and user equipment |
| CN115529201A (en) * | 2022-05-31 | 2022-12-27 | 青岛海尔智能家电科技有限公司 | Method, system, device, server and storage medium for generating family environment panorama based on digital twinning |
| CN116206325A (en) * | 2023-02-14 | 2023-06-02 | 南京艾小宝智能科技有限公司 | System and method for identifying house type graph |
| CN116311325A (en) * | 2023-02-16 | 2023-06-23 | 江苏艾佳家居用品有限公司 | Automatic scale identification system based on artificial intelligence model |
Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN101571875A (en) * | 2009-05-05 | 2009-11-04 | 程治永 | Realization method of image searching system based on image recognition |
| CN104217206A (en) * | 2013-05-31 | 2014-12-17 | 上海亚视信息科技有限公司 | Real-time attendance counting method based on high-definition videos |
| CN105701502A (en) * | 2016-01-06 | 2016-06-22 | 福州大学 | Image automatic marking method based on Monte Carlo data balance |
| CN106023220A (en) * | 2016-05-26 | 2016-10-12 | 史方 | Vehicle exterior part image segmentation method based on deep learning |
-
2017
- 2017-01-18 CN CN201710036469.4A patent/CN106844614A/en active Pending
Patent Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN101571875A (en) * | 2009-05-05 | 2009-11-04 | 程治永 | Realization method of image searching system based on image recognition |
| CN104217206A (en) * | 2013-05-31 | 2014-12-17 | 上海亚视信息科技有限公司 | Real-time attendance counting method based on high-definition videos |
| CN105701502A (en) * | 2016-01-06 | 2016-06-22 | 福州大学 | Image automatic marking method based on Monte Carlo data balance |
| CN106023220A (en) * | 2016-05-26 | 2016-10-12 | 史方 | Vehicle exterior part image segmentation method based on deep learning |
Cited By (33)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN107545306A (en) * | 2017-07-05 | 2018-01-05 | 安徽奇智科技有限公司 | A kind of big data analysis platform based on cloud computing |
| CN108009598A (en) * | 2017-12-27 | 2018-05-08 | 北京诸葛找房信息技术有限公司 | Floor plan recognition methods based on deep learning |
| CN108764022A (en) * | 2018-04-04 | 2018-11-06 | 链家网（北京）科技有限公司 | A kind of image-recognizing method and system |
| CN108268748A (en) * | 2018-04-26 | 2018-07-10 | 南京信息工程大学 | A kind of residence model figure electrical design method based on machine learning |
| CN108984765B (en) * | 2018-07-20 | 2020-01-31 | 我要家网络科技有限公司 | family pattern house finding method based on intelligent image analysis |
| CN108984765A (en) * | 2018-07-20 | 2018-12-11 | 我要家网络科技有限公司 | It is a kind of based on intelligent image analysis floor plan look for room method |
| CN109711443A (en) * | 2018-12-14 | 2019-05-03 | 平安城市建设科技(深圳)有限公司 | Floor plan recognition methods, device, equipment and storage medium neural network based |
| CN109740506A (en) * | 2018-12-29 | 2019-05-10 | 广东三维家信息科技有限公司 | A kind of house type image-recognizing method and device |
| CN110059721A (en) * | 2019-03-16 | 2019-07-26 | 平安城市建设科技(深圳)有限公司 | Floor plan area recognizing method, device, equipment and computer readable storage medium |
| CN110059690A (en) * | 2019-03-28 | 2019-07-26 | 广州智方信息科技有限公司 | Floor plan semanteme automatic analysis method and system based on depth convolutional neural networks |
| CN109993797B (en) * | 2019-04-04 | 2021-03-02 | 广东三维家信息科技有限公司 | Door and window position detection method and device |
| CN109993797A (en) * | 2019-04-04 | 2019-07-09 | 广东三维家信息科技有限公司 | Door and window method for detecting position and device |
| CN110059750A (en) * | 2019-04-17 | 2019-07-26 | 广东三维家信息科技有限公司 | House type shape recognition process, device and equipment |
| CN110111426A (en) * | 2019-04-18 | 2019-08-09 | 贝壳技术有限公司 | A kind of determination method and apparatus in sound separate pattern house |
| CN110197225A (en) * | 2019-05-28 | 2019-09-03 | 广东三维家信息科技有限公司 | House type spatial match method and system based on deep learning |
| CN110443895B (en) * | 2019-06-27 | 2020-08-18 | 贝壳找房(北京)科技有限公司 | House lighting scheme generation method and device |
| CN110443895A (en) * | 2019-06-27 | 2019-11-12 | 贝壳技术有限公司 | House illumination scheme generation method and device |
| CN110390153A (en) * | 2019-07-15 | 2019-10-29 | 贝壳技术有限公司 | Generation method, device and the equipment of layout structure modification scheme, storage medium |
| CN110390304A (en) * | 2019-07-24 | 2019-10-29 | 广东南方数码科技股份有限公司 | Automatic classification method, device, electronic equipment and storage medium |
| CN110728235A (en) * | 2019-10-11 | 2020-01-24 | 广东三维家信息科技有限公司 | House type area marking method and device |
| CN110826627A (en) * | 2019-11-06 | 2020-02-21 | 广东三维家信息科技有限公司 | Image similarity measuring method and device and electronic equipment |
| CN111400519B (en) * | 2019-12-05 | 2022-08-12 | 江苏艾佳家居用品有限公司 | Efficient customized massive similar house type searching method |
| CN111400519A (en) * | 2019-12-05 | 2020-07-10 | 江苏艾佳家居用品有限公司 | Efficient customized massive similar house type searching method |
| CN111291649A (en) * | 2020-01-20 | 2020-06-16 | 广东三维家信息科技有限公司 | Image recognition method and device and electronic equipment |
| CN111291649B (en) * | 2020-01-20 | 2023-08-25 | 广东三维家信息科技有限公司 | Image recognition method and device and electronic equipment |
| CN111798340A (en) * | 2020-06-24 | 2020-10-20 | 北京字节跳动网络技术有限公司 | House type graph identification method and device and electronic equipment |
| CN112116613A (en) * | 2020-09-25 | 2020-12-22 | 贝壳技术有限公司 | Model training method, image segmentation method, image vectorization method and system thereof |
| CN114554503A (en) * | 2020-11-24 | 2022-05-27 | 中移物联网有限公司 | Method and device for confirming networking information and user equipment |
| CN114554503B (en) * | 2020-11-24 | 2023-11-10 | 中移物联网有限公司 | Networking information confirmation method and device and user equipment |
| CN115529201A (en) * | 2022-05-31 | 2022-12-27 | 青岛海尔智能家电科技有限公司 | Method, system, device, server and storage medium for generating family environment panorama based on digital twinning |
| CN116206325A (en) * | 2023-02-14 | 2023-06-02 | 南京艾小宝智能科技有限公司 | System and method for identifying house type graph |
| CN116311325A (en) * | 2023-02-16 | 2023-06-23 | 江苏艾佳家居用品有限公司 | Automatic scale identification system based on artificial intelligence model |
| CN116311325B (en) * | 2023-02-16 | 2023-10-27 | 江苏艾佳家居用品有限公司 | Automatic scale identification system based on artificial intelligence model |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN106844614A (en) | A kind of floor plan functional area system for rapidly identifying | |
| CN106971152B (en) | Method for detecting bird nest in power transmission line based on aerial images | |
| CN104992142B (en) | A kind of pedestrian recognition method being combined based on deep learning and attribute study | |
| CN110222611A (en) | Human skeleton Activity recognition method, system, device based on figure convolutional network | |
| CN108416314B (en) | Picture important face detection method | |
| CN105938564A (en) | Rice disease recognition method based on principal component analysis and neural network and rice disease recognition system thereof | |
| CN109543679A (en) | A kind of dead fish recognition methods and early warning system based on depth convolutional neural networks | |
| CN109978918A (en) | A kind of trajectory track method, apparatus and storage medium | |
| CN110245720B (en) | Deep learning-based intelligent citrus pest diagnosis method and system | |
| CN104217214A (en) | Configurable convolutional neural network based red green blue-distance (RGB-D) figure behavior identification method | |
| CN107766582A (en) | A kind of image search method based on target regional area | |
| CN110139067A (en) | A kind of wild animal monitoring data management information system | |
| CN109344699A (en) | Winter jujube disease recognition method based on depth of seam division convolutional neural networks | |
| Bertrand et al. | Bark and leaf fusion systems to improve automatic tree species recognition | |
| CN105590099A (en) | Multi-user behavior identification method based on improved convolutional neural network | |
| CN110458082A (en) | A kind of city management case classification recognition methods | |
| CN104680193B (en) | Online objective classification method and system based on quick similitude network integration algorithm | |
| CN112668375B (en) | Tourist distribution analysis system and method in scenic spot | |
| CN106845513A (en) | Staff detector and method based on condition random forest | |
| CN114359727A (en) | Tea disease identification method and system based on lightweight optimization Yolo v4 | |
| CN108549901A (en) | A kind of iteratively faster object detection method based on deep learning | |
| Hsieh et al. | Fruit maturity and location identification of beef tomato using R-CNN and binocular imaging technology | |
| CN111046213B (en) | Knowledge base construction method based on image recognition | |
| CN109992781A (en) | Processing, device, storage medium and the processor of text feature | |
| CN107886077A (en) | A kind of crop pests recognition methods and its system based on wechat public number |
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 | ||
| WD01 | Invention patent application deemed withdrawn after publication | ||
| WD01 | Invention patent application deemed withdrawn after publication |
Application publication date: 20170613 |