CN110096609A - Source of houses searching method, device, equipment and computer readable storage medium - Google Patents

Source of houses searching method, device, equipment and computer readable storage medium Download PDF

Info

Publication number
CN110096609A
CN110096609A CN201910202677.6A CN201910202677A CN110096609A CN 110096609 A CN110096609 A CN 110096609A CN 201910202677 A CN201910202677 A CN 201910202677A CN 110096609 A CN110096609 A CN 110096609A
Authority
CN
China
Prior art keywords
source
houses
picture
sample
recommendation
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
Application number
CN201910202677.6A
Other languages
Chinese (zh)
Inventor
徐伟
王威
何维
曾文韬
陈佳
杨昭
朱世新
Current Assignee (The listed assignees may be inaccurate. Google has not performed a legal analysis and makes no representation or warranty as to the accuracy of the list.)
Ping An Urban Construction Technology Shenzhen Co Ltd
Original Assignee
Ping An Urban Construction Technology Shenzhen Co Ltd
Priority date (The priority date 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 date listed.)
Filing date
Publication date
Application filed by Ping An Urban Construction Technology Shenzhen Co Ltd filed Critical Ping An Urban Construction Technology Shenzhen Co Ltd
Priority to CN201910202677.6A priority Critical patent/CN110096609A/en
Publication of CN110096609A publication Critical patent/CN110096609A/en
Pending legal-status Critical Current

Links

Classifications

    • GPHYSICS
    • G06COMPUTING; CALCULATING OR COUNTING
    • G06FELECTRIC DIGITAL DATA PROCESSING
    • G06F16/00Information retrieval; Database structures therefor; File system structures therefor
    • G06F16/50Information retrieval; Database structures therefor; File system structures therefor of still image data
    • G06F16/58Retrieval characterised by using metadata, e.g. metadata not derived from the content or metadata generated manually
    • G06F16/583Retrieval characterised by using metadata, e.g. metadata not derived from the content or metadata generated manually using metadata automatically derived from the content
    • GPHYSICS
    • G06COMPUTING; CALCULATING OR COUNTING
    • G06QINFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
    • G06Q50/00Information and communication technology [ICT] specially adapted for implementation of business processes of specific business sectors, e.g. utilities or tourism
    • G06Q50/10Services
    • G06Q50/16Real estate

Landscapes

  • Engineering & Computer Science (AREA)
  • Theoretical Computer Science (AREA)
  • Business, Economics & Management (AREA)
  • Physics & Mathematics (AREA)
  • Tourism & Hospitality (AREA)
  • Library & Information Science (AREA)
  • General Physics & Mathematics (AREA)
  • General Health & Medical Sciences (AREA)
  • Strategic Management (AREA)
  • Primary Health Care (AREA)
  • General Business, Economics & Management (AREA)
  • Marketing (AREA)
  • Human Resources & Organizations (AREA)
  • Economics (AREA)
  • Health & Medical Sciences (AREA)
  • Data Mining & Analysis (AREA)
  • Databases & Information Systems (AREA)
  • General Engineering & Computer Science (AREA)
  • Information Retrieval, Db Structures And Fs Structures Therefor (AREA)

Abstract

The present invention relates to field of image detection, a kind of source of houses searching method, device, equipment and computer readable storage medium are provided, this method comprises: obtaining demand source of houses picture according to the source of houses search instruction when receiving the source of houses search instruction of search terminal transmission;Example segmentation is carried out to the demand source of houses picture based on default parted pattern, obtain source of houses main body example and non-source of houses main body example, and feature extraction is carried out to the source of houses main body example and the non-source of houses main body example respectively, obtain corresponding source of houses body feature and non-source of houses body feature;Source of houses retrieval is carried out in default source of houses library according to the source of houses body feature and the non-source of houses body feature, determines the recommendation source of houses corresponding with the demand source of houses picture;The recommendation information of real estate for recommending the source of houses is obtained, and the recommendation information of real estate is sent to described search terminal.The present invention can receive user and provide source of houses picture, and realize to scheme to search room, improve the usage experience of user.

Description

Source of houses searching method, device, equipment and computer readable storage medium
Technical field
The present invention relates to field of image detection more particularly to a kind of source of houses searching method, device, equipment and computer-readable Storage medium.
Background technique
With the continuous development of society, the side that the service selling room Yu renting a house is passed through internet by more and more house property mediums Formula is promoted, and when tenant or house purchaser are intended to search wait lease or when house to be vended, can pass through house property medium operation House property site search information of real estate and rented a house or bought house.
User is in the enterprising having sexual intercourse source information search in house property medium website at present, it usually needs relies on source of houses address, price Etc. factors carry out source of houses search;But when user only has the picture of some source of houses can not be in house property when not knowing the title of the source of houses but The information of real estate is found when intermediary web site, this is just that user brings inconvenience.
Summary of the invention
The main purpose of the present invention is to provide a kind of source of houses searching method, device, equipment and computer-readable storage mediums Matter, it is intended to which solution can not be by source of houses picture come the technical issues of searching for the source of houses.
To achieve the above object, the present invention provides a kind of source of houses searching method, and the source of houses searching method includes:
When receiving the source of houses search instruction of search terminal transmission, the demand source of houses is obtained according to the source of houses search instruction Picture;
Example segmentation is carried out to the demand source of houses picture based on default parted pattern, obtains source of houses main body example and non-room Source main body example, and feature extraction is carried out to the source of houses main body example and the non-source of houses main body example respectively, it is corresponded to Source of houses body feature and non-source of houses body feature;
Source of houses retrieval is carried out in default source of houses library according to the source of houses body feature and the non-source of houses body feature, really The fixed recommendation source of houses corresponding with the demand source of houses picture;
The recommendation information of real estate for recommending the source of houses is obtained, and the recommendation information of real estate is sent to described search end End.
In addition, to achieve the above object, the present invention also provides a kind of source of houses searcher, the source of houses searcher packet It includes:
Picture obtains module, for being searched according to the source of houses when receiving the source of houses search instruction of search terminal transmission Suo Zhiling obtains demand source of houses picture;
Characteristic extracting module is obtained for carrying out example segmentation to the demand source of houses picture based on default parted pattern Source of houses main body example and non-source of houses main body example, and the source of houses main body example and the non-source of houses main body example are carried out respectively Feature extraction obtains corresponding source of houses body feature and non-source of houses body feature;
Source of houses retrieval module is used for according to the source of houses body feature and the non-source of houses body feature in default source of houses library Middle progress source of houses retrieval determines the recommendation source of houses corresponding with the demand source of houses picture;
First sending module, for obtaining the recommendation information of real estate for recommending the source of houses, and by the recommendation information of real estate It is sent to described search terminal.
In addition, to achieve the above object, the present invention also provides a kind of sources of houses to search for equipment, the source of houses search equipment includes Processor, memory and it is stored in the source of houses search program that can be executed on the memory and by the processor, wherein institute When stating source of houses search program and being executed by the processor, realize such as the step of above-mentioned source of houses searching method.
In addition, to achieve the above object, the present invention also provides a kind of storage medium, being stored with the source of houses on the storage medium Search program, wherein realizing when the source of houses search program is executed by processor such as the step of above-mentioned source of houses searching method.
The present invention can receive the source of houses picture that user provides, and carries out feature to the source of houses picture from different dimensions and mention It takes, and source of houses search is carried out in known source of houses library according to the feature of extraction, associated recommendation information of real estate is finally returned that, to solve It certainly currently has to input specific source of houses position or title just can be carried out the pain spot of source of houses retrieval, user is facilitated not know the source of houses Source of houses search is carried out when specifying information, improves the usage experience of user.
Detailed description of the invention
Fig. 1 is the hardware structural diagram that the source of houses involved in the embodiment of the present invention searches for equipment;
Fig. 2 is the flow diagram of source of houses searching method first embodiment of the present invention;
Fig. 3 is the functional block diagram of source of houses searcher first embodiment of the present invention.
The embodiments will be further described with reference to the accompanying drawings for the realization, the function and the advantages of the object of the present invention.
Specific embodiment
It should be appreciated that the specific embodiments described herein are merely illustrative of the present invention, it is not intended to limit the present invention.
The present embodiments relate to source of houses searching method be mainly used in the source of houses search equipment, the source of houses search for equipment can Being realized by equipment having data processing function such as personal computer (personal computer, PC), servers.
Referring to Fig.1, Fig. 1 is the hardware structural diagram that the source of houses involved in the embodiment of the present invention searches for equipment.This In inventive embodiments, source of houses search equipment may include (such as the central processing unit Central of processor 1001 Processing Unit, CPU), communication bus 1002, user interface 1003, network interface 1004, memory 1005.Wherein, Communication bus 1002 is for realizing the connection communication between these components;User interface 1003 may include display screen (Display), input unit such as keyboard (Keyboard);Network interface 1004 optionally may include that the wired of standard connects Mouth, wireless interface (such as Wireless Fidelity WIreless-FIdelity, WI-FI interface);Memory 1005 can be high speed and deposit at random Access to memory (random access memory, RAM), is also possible to stable memory (non-volatile memory), Such as magnetic disk storage, memory 1005 optionally can also be the storage device independently of aforementioned processor 1001.This field Technical staff is appreciated that hardware configuration shown in Fig. 1 and does not constitute a limitation of the invention, and may include more than illustrating Or less component, perhaps combine certain components or different component layouts.
With continued reference to Fig. 1, the memory 1005 in Fig. 1 as a kind of computer readable storage medium may include operation system System, network communication module and source of houses search program.In Fig. 1, network communication module is mainly used for connecting database, with data Library carries out data communication;And processor 1001 can call the source of houses search program stored in memory 1005, and execute this hair The source of houses searching method that bright embodiment provides.
The embodiment of the invention provides a kind of source of houses searching methods.
It is the flow diagram of source of houses searching method first embodiment of the present invention referring to Fig. 2, Fig. 2.
In the present embodiment, the source of houses searching method the following steps are included:
Step S10 is obtained when receiving the source of houses search instruction of search terminal transmission according to the source of houses search instruction Demand source of houses picture;
With the continuous development of society, the side that the service selling room Yu renting a house is passed through internet by more and more house property mediums Formula is promoted, and when tenant or house purchaser are intended to search wait lease or when house to be vended, can pass through house property medium operation House property site search information of real estate and rented a house or bought house.User searches in the enterprising having sexual intercourse source information in house property medium website at present Suo Shi, it usually needs carry out source of houses search by factors such as source of houses address, prices;But work as the picture that user only has some source of houses, When not knowing the title of the source of houses but, the information of real estate can not be found at house property medium website, this is just that user brings not Just.In this regard, propose a kind of source of houses searching method in the present embodiment, it can receive user and source of houses picture be provided, and from different dimensions Feature extraction is carried out to the source of houses picture, and source of houses search is carried out in known source of houses library according to the feature of extraction, is finally returned that Associated recommendation information of real estate, so that solution currently has to input specific source of houses position or title just can be carried out the pain of source of houses retrieval Point facilitates user to carry out source of houses search when not knowing source of houses specifying information, improves the usage experience of user.
The source of houses searching method of the present embodiment be by the source of houses search equipment realize, the equipment by taking search server as an example into Row explanation.In the present embodiment, when user (such as client, staff) needs to search for corresponding source of houses letter by way of picture It when breath, can be operated in search terminal, from the local source of houses picture for uploading or specifying certain to throw the net on network as demand source of houses figure Piece, and trigger source of houses searching request;Wherein, which can be personal computer PC, mobile phone, tablet computer etc..Search Terminal can send corresponding source of houses search instruction to search server, which searches when receiving the source of houses searching request It include the network storage address from the demand source of houses picture locally uploaded or the demand source of houses picture in Suo Zhiling.When search takes When business device receives the source of houses search instruction of search terminal transmission, the need for needing to identify can be obtained according to the source of houses search instruction Seek source of houses picture.
It is worth noting that in order to further increase the efficiency and source of houses effectiveness of retrieval of subsequent picture processing, it is right In the source of houses picture that user uploads, the picture attributes such as its photo resolution, dimension of picture, file format can be required; Search server first detects whether the demand source of houses picture meets the requirements when getting demand source of houses picture, if not satisfied, then It can be returned to search terminal and upload prompt again, to prompt user to upload source of houses picture again;If satisfied, then carrying out next processing Step.
Step S20 carries out example segmentation to the demand source of houses picture based on default parted pattern, it is real to obtain source of houses main body Example and non-source of houses main body example, and feature extraction is carried out to the source of houses main body example and the non-source of houses main body example respectively, Obtain corresponding source of houses body feature and non-source of houses body feature;
In the present embodiment, search server, can be according to the actual situation to the demand when getting the demand source of houses picture Source of houses picture is pre-processed, such as the unification of dimension of picture, pixel value normalize (such as linear function normalization, logarithmic function Normalization) etc., naturally it is also possible to without pretreatment;Then will to the demand source of houses picture carry out feature extraction, and according to This feature carries out source of houses retrieval.It is worth noting that carrying out source of houses retrieval in the present embodiment, it is contemplated that different house properties, Neighbouring natural environment, building etc. also can different from, and the features such as these natural environments, building also can be used as the source of houses retrieval One clue.In this regard, search server in the present embodiment first will based on a default parted pattern to demand source of houses picture into The segmentation of row example, obtains source of houses main body example and non-source of houses main body example;Example segmentation is target identification (object Detection it) is combined with one of semantic segmentation (semantic segmentation);Wherein target identification is to detect one There is any target in width image, while the location position of target being come out, and semantic segmentation is by each pixel in an image Fixed classification delimited, that is, classified to each pixel;Source of houses main body example in the present embodiment, the as source of houses (house property) main part, non-source of houses main body example includes example environments (such as trees, lawn), (such as next door is small for adjacent building example Area, electric pole) etc..Certainly, in practice, the demand source of houses picture uploaded for user, may only include one of above-mentioned Non- source of houses main body example, such as only include trees.
Default parted pattern in the present embodiment can be and be obtained by way of machine learning (training);Certainly should Default parted pattern can be different method and be constructed, and the default parted pattern in the present embodiment is based on masked areas- Convolutional neural networks Mask R-CNN building, Mask R-CNN is in fast area-convolutional neural networks Faster R-CNN On the basis of be added to an exposure mask mask branch, semantic segmentation (semantic is done to each interest region (ROI) segmentation).Specifically, search server first will be by presetting the predeterminable area candidate network RPN in parted pattern (Region Proposal Network) detects demand source of houses picture, determines the region of interest in the demand source of houses picture Domain ROI (Region of Interests), interest region ROI are source of houses main body example and non-source of houses main body example place Region.Meanwhile search server will also carry out process of convolution to demand source of houses picture by default convolutional layer, obtain to extract The global feature figure of the demand source of houses picture;In view of the default parted pattern in the present embodiment is constructed based on Mask R-CNN , and Mask R-CNN is to obtain on the basis of Faster R-CNN, therefore the default convolutional layer can be direct borrow Convolutional layer in Faster R-CNN.In the interest worlds ROI and global feature figure for obtaining demand source of houses picture, according to interest Distribution of the field in the demand source of houses picture can extract from the global feature figure and obtain corresponding interest provincial characteristics Figure.It is worth noting that interest provincial characteristics figure at this time can be the characteristic pattern of multiple regions.Then, search server can Pixel correction is carried out to each interest provincial characteristics figure, the process of the pixel correction can be through ROIAlign realization, ROIAlign eliminates quantization operation compared with ROI Pooling, and obtaining coordinate using the method for bilinear interpolation is floating-point Image values on several pixels, to convert a continuous operation for entire feature accumulation process.It is completed in correction When, it can be special to the interest region after correction by presetting full convolutional network FCN (Fully Convolutional Networks) Sign figure is analyzed, to determine the example types in interest region;It, can be according to this when determining the example types in each interest region Example types predict each each pixel generic of ROI, to be split to demand source of houses picture, it is real to obtain source of houses main body Example and non-source of houses main body example.
In the present embodiment, when obtaining source of houses main body example and non-source of houses main body example, search server will be respectively to room Source main body example and non-source of houses main body example carry out feature extraction, obtain corresponding source of houses body feature and non-source of houses main body is special Sign.For the process that this feature is extracted, it can be and realized by trained machine mould in advance, not such as neural network, scale Become eigentransformation sift technology etc.;And for above-mentioned feature, can be is indicated in a manner of vector, naturally it is also possible to be with word The mode of symbol string indicates.
Step S30 carries out room in default source of houses library according to the source of houses body feature and the non-source of houses body feature Source retrieval, determines the recommendation source of houses corresponding with the demand source of houses picture;
It, can be special according to the source of houses main body when obtaining source of houses body feature and non-source of houses body feature in the present embodiment Non- source of houses body feature of seeking peace carries out source of houses retrieval in default source of houses library, searches the sample source of houses with same characteristic features, and The recommendation source of houses corresponding with demand source of houses picture is determined in these sample sources of houses.It wherein, include preparatory in the default source of houses library Several sample sources of houses collected, further include the sample source of houses picture for having these sample sources of houses, further include having these sample source of houses figures Samples pictures feature corresponding to piece;These samples pictures features are also by the way of such as above-mentioned steps S20 from the sample source of houses It extracts and obtains in picture, including sample body feature and non-sample main body.The process of source of houses retrieval, namely it is based on the demand source of houses The sample body feature and non-sample main body of the source of houses main body example of picture and non-source of houses main body example and sample source of houses picture It carries out.Specifically, step S30 includes:
The sample source of houses picture of each sample source of houses is obtained in default source of houses library, and obtains the sample of the sample source of houses picture This body feature and non-sample body feature;
Search server can obtain the sample source of houses picture and these samples of each sample source of houses from default source of houses library first The sample body feature and non-sample body feature of this source of houses picture.
The sample body feature and the source of houses body feature are compared, determine the demand source of houses picture and sample The same body characteristic of this source of houses picture, and the non-sample body feature and the non-source of houses body feature are compared It is right, determine demand source of houses picture nonbody characteristic identical as sample source of houses picture;
It, can be by each sample source of houses figure in the sample body feature and non-sample body feature for obtaining each sample source of houses picture The sample body feature of piece compared with the source of houses body feature of demand source of houses picture, determines demand source of houses picture and sample room respectively The same body characteristic of source picture;And by the non-source of houses master of the non-sample body feature of each sample source of houses and demand source of houses picture Body characteristics are compared, and determine demand source of houses picture nonbody characteristic identical as sample source of houses picture.For example, demand source of houses picture Same body characteristic with A sample source of houses picture is x1, is x2 with the identical nonbody characteristic of A sample source of houses picture;It needs Seeking the same body characteristic of source of houses picture and B sample source of houses picture is y1, with the identical non-master body characteristics of B sample source of houses picture Number is y2.Wherein, during judging whether demand source of houses picture is identical with the feature of sample source of houses picture, it can be and pass through The characteristic distance of two features is calculated to judge, for example, when two features be vector form when, can be calculate Euclidean distance, Mahalanobis distance Mahalanobis Distance etc. can be calculating Hamming distance when two features are character string forms;When When the distance of two features is less than a preset threshold, i.e., it is believed that the two be characterized in it is identical.
According to the same body characteristic and the identical nonbody characteristic in each sample source of houses determine with The corresponding recommendation source of houses of the demand source of houses picture.
In the same body characteristic and identical nonbody characteristic for determining each sample source of houses picture, it can according to this Same body characteristic and identical nonbody characteristic determine in each sample source of houses recommends the source of houses.Such as it can be and calculate separately The feature summation of same body characteristic and identical nonbody characteristic in every sample source of houses picture, and feature summation is most Several sample source of houses pictures as recommendation source of houses picture corresponding with demand source of houses picture, corresponding to the recommendation source of houses picture The sample source of houses be recommend the source of houses.Certainly, in practice, the body feature in source of houses picture and non-master body characteristics are to identification one The significance level that a source of houses rises may be different, therefore can also be the determining recommendation source of houses in weighted fashion.Specifically, For every sample source of houses picture, its corresponding same body characteristic and identical nonbody characteristic can be substituting to default phase Seemingly divide in formula, with similar point of picture of the weighted calculation sample source of houses picture, wherein presetting similar point of formula are as follows:
P=k1*N1+k2*N2
Wherein, P is similar point of picture of the sample source of houses picture;
K1 is the first default feature specific gravity, k1 > 0;
K2 is the second default feature specific gravity, k2 > 0;
N1 is same body characteristic corresponding to the sample source of houses picture, N1 >=0;
N2 is same body characteristic corresponding to the sample source of houses picture, N2 >=0;
In the similar timesharing of picture that every sample source of houses picture is calculated, can by similar point of picture it is highest several Sample source of houses picture is as recommendation source of houses picture corresponding with demand source of houses picture, sample room corresponding to the recommendation source of houses picture Source is to recommend the source of houses.
Step S40 obtains the recommendation information of real estate for recommending the source of houses, and described in the recommendations information of real estate is sent to Search terminal.
In the present embodiment, determining that search server will obtain the recommendation source of houses letter of the recommendation source of houses when recommending the source of houses Breath, the recommendation information of real estate include recommending source of houses affiliated subdistrict, the more pictures of cell, cell address, cell the built time, small Area's source of houses room rate, cordon traffic information etc..Information of real estate is recommended for these, can be and be stored in advance in above-mentioned default source of houses library In, it is also possible to search server and is inquired and obtained from network again after determining the recommendation source of houses.Server is being somebody's turn to do When recommending information of real estate, which can be sent in search terminal, be somebody's turn to do so that user is understood by search terminal The source of houses (cell) information.
Certainly, it is determining that the recommendation source of houses may be more than two sources of houses when recommending the source of houses, thus is recommending room obtaining It is to obtain more than two recommendation informations of real estate for recommending the source of houses when source information.At this point, in order to facilitate user's quick obtaining oneself The information wanted to know about, search server can recommend the source of houses to be ranked up these based on a predetermined order rule, then will sequence Recommendation information of real estate afterwards is sent to search terminal.Wherein the predetermined order rule can be is configured according to the actual situation, Optionally the ordering rule can be the source of houses retrieval hits for recommending the source of houses in default measurement period according to these and (be retrieved To number) be configured sequence, such as recommend the source of houses and N to recommend the source of houses M, M recommends the source of houses in upper retrieval in one month Hits are 100 times, and the N recommendation source of houses is 90 times in upper one month retrieval hits, these retrieval hits can be positively correlated Mode be quantified as corresponding sequence point, retrieval hit is point higher, and sequence point is higher, then sort more forward namely above-mentioned row Sequence can be for by M to N.And after these recommendation informations of real estate are sent to search terminal, search terminal also will acquire search eventually Hold to these recommend informations of real estate operation note, such as user click which recommend the source of houses, to recommend the source of houses browsing when Grow, whether carry out collection sharing operation, recommendation results are selected with satisfaction or is unsatisfied with etc.;It would know that use by these operation notes Level of interest of source of houses is recommended these at family, and is adjusted according to these operation notes to above-mentioned ordering rule, such as root Operation note is quantified as the corresponding interest for recommending the source of houses point according to a certain amount rule, then divides and sorts by the interest Divide after being weighted and reset according to weighted results, so that sequence is closer to actual conditions.
Further, in the present embodiment, after sending recommendation information of real estate to search terminal, know more in order to facilitate user More informations of real estate, search server can also push more informations of real estate to user according to the characteristics of recommending information of real estate. Specifically, further including having to recommend source of houses address and/or recommend source of houses room rate in the recommendation information of real estate that search server obtains, search Rope server can obtain matched with the recommendation source of houses according to the recommendation source of houses address and/or recommendation source of houses room rate of the recommendation source of houses The similar source of houses, such as will be 1km at a distance from the recommendation source of houses centered on the recommendation source of houses, the progress source of houses search of 1km radius Within the source of houses as the similar source of houses;In another example it (is such as all 10000-12000 that search, which has identical room rate range with the recommendation source of houses, Member/square meter) the source of houses as the similar source of houses.Then search server will acquire the similar information of real estate of the similar source of houses, and will The similar source of houses is sent to search terminal.
In the present embodiment, when receiving the source of houses search instruction of search terminal transmission, according to the source of houses search instruction Acquisition demand source of houses picture;Example segmentation is carried out to the demand source of houses picture based on default parted pattern, obtains source of houses main body Example and non-source of houses main body example, and feature is carried out to the source of houses main body example and the non-source of houses main body example respectively and is mentioned It takes, obtains corresponding source of houses body feature and non-source of houses body feature;According to the source of houses body feature and the non-source of houses master Body characteristics carry out source of houses retrieval in default source of houses library, determine the recommendation source of houses corresponding with the demand source of houses picture;Obtain institute The recommendation information of real estate for recommending the source of houses is stated, and the recommendation information of real estate is sent to described search terminal.In the above manner, A kind of source of houses searching method is proposed in the present embodiment, can receive the source of houses picture that user provides, and from different dimensions to the room Source picture carries out feature extraction, and source of houses search is carried out in known source of houses library according to the feature of extraction, finally returns that correlation pushes away Information of real estate is recommended, so that solution currently has to input specific source of houses position or title just can be carried out the pain spot of source of houses retrieval, side Just user carries out source of houses search when not knowing source of houses specifying information, improves the usage experience of user.
Based on above-mentioned embodiment illustrated in fig. 2, source of houses searching method second embodiment of the present invention is proposed.
In the present embodiment, after step S40 further include:
When receiving the navigation instruction of described search terminal transmission, according to the navigation instruction in the recommendation source of houses It determines the target source of houses, and obtains the destination address of the target source of houses and the terminal address of described search terminal;
In the present embodiment, after sending recommendation information of real estate to search terminal, its intention is known in order to facilitate recommended user The place of the source of houses and the method for reaching the source of houses, search server can also provide navigation feature for user.Specifically, search clothes Business device is sent in the recommendation information of real estate of search terminal, further includes having navigation options;User pushes away being checked by search terminal When recommending information of real estate, if interested in a certain recommendation source of houses, the navigation options can be clicked;Search terminal is then depending on the user's operation Corresponding navigation instruction is sent to search server.Search server refers to when receiving the navigation instruction according to the navigation It enables and determines the target source of houses in recommending the source of houses, and obtain the destination address where the target source of houses;Server is simultaneously scanned for will also The current terminal address of search terminal is obtained, which is regarded as the current location of user.For the terminal address, It can be search terminal to collect by itself GPS device, and sent together when sending navigation instruction to search server, It is also possible to search server to be positioned to obtain according to the IP address of search terminal, can also be that other modes obtain certainly Obtain (such as user fills in manually).
Navigation routine is generated according to the destination address and the terminal address, and navigation routine is back to described search Terminal.
In the present embodiment, search server when obtaining destination address and terminal address, can according to the destination address and Terminal address generates navigation routine, and the navigation routine is back to search terminal, so that user obtains according to the navigation routine The method for reaching the target source of houses.Certain navigation routine can be including a variety of trip modes and a variety of traffic paths, such as Driving route, bus routes, Walking Route etc. can also include each trip mode and the estimated trip time-consuming of traffic path etc.. In the above manner, navigation feature can be provided for user, facilitates recommended user to know the place of its intention source of houses and reach and be somebody's turn to do The method of the source of houses is conducive to the usage experience for improving user.
In addition, the embodiment of the present invention also provides a kind of source of houses searcher.
It is the functional block diagram of source of houses searcher first embodiment of the present invention referring to Fig. 3, Fig. 3.
In the present embodiment, the source of houses searcher includes:
Picture obtain module 10, for receive search terminal transmission the source of houses search instruction when, according to the source of houses Search instruction obtains demand source of houses picture;
Characteristic extracting module 20 is obtained for carrying out example segmentation to the demand source of houses picture based on default parted pattern Source of houses main body example and non-source of houses main body example, and respectively to the source of houses main body example and the non-source of houses main body example into Row feature extraction obtains corresponding source of houses body feature and non-source of houses body feature;
Source of houses retrieval module 30 is used for according to the source of houses body feature and the non-source of houses body feature in the default source of houses Source of houses retrieval is carried out in library, determines the recommendation source of houses corresponding with the demand source of houses picture;
First sending module 40 is believed for obtaining the recommendation information of real estate for recommending the source of houses, and by the recommendation source of houses Breath is sent to described search terminal.
Wherein, each virtual functions module of above-mentioned source of houses searcher is stored in the storage of the search equipment of the source of houses shown in Fig. 1 It is functional for realizing the institute of source of houses search program in device 1005;When each module is executed by processor 1001, it can be achieved that based on searching The source of houses search instruction that rope terminal is sent is carried out to scheme to search the function in room.
Further, the characteristic extracting module 20 includes:
Picture detection unit, for being detected by predeterminable area candidate network RPN to the demand source of houses picture, really The interest region of the fixed demand source of houses picture;
Convolution processing unit obtains whole for carrying out process of convolution to the demand source of houses picture by default convolutional layer Body characteristics figure, and extracted from the global feature figure according to the interest region and obtain corresponding interest provincial characteristics figure;
Type determining units, for carrying out pixel correction to the interest provincial characteristics figure, and by presetting full convolution net Network FCN analyzes the interest provincial characteristics figure after correction, with the example types in the determination interest region;
Picture genre cell, for being divided according to the example types in the interest region the demand source of houses picture It cuts, obtains source of houses main body example and non-source of houses main body example.
Further, the source of houses retrieval module 30 includes:
Sample acquisition unit for obtaining the sample source of houses picture of each sample source of houses in default source of houses library, and obtains institute State the sample body feature and non-sample body feature of sample source of houses picture;
Feature comparing unit determines institute for comparing the sample body feature and the source of houses body feature State the same body characteristic of demand source of houses picture Yu the sample source of houses picture, and by the non-sample body feature and described Non- source of houses body feature compares, and determines the demand source of houses picture non-master body characteristics identical as the sample source of houses picture Number;
Source of houses determination unit is used for according to the same body characteristic and the identical nonbody characteristic described each The recommendation source of houses corresponding with the demand source of houses picture is determined in the sample source of houses.
Further, the source of houses determination unit is specifically used for the same body characteristic and identical nonbody is special Sign number, which is substituting to, to be preset in similar point of formula, to calculate similar point of picture of the sample source of houses picture, wherein the default phase Seemingly divide formula are as follows:
P=k1*N1+k2*N2
Wherein, P is similar point of picture of the sample source of houses picture;
K1 is the first default feature specific gravity, k1 > 0;
K2 is the second default feature specific gravity, k2 > 0;
N1 is same body characteristic corresponding to the sample source of houses picture, N1 >=0;
N2 is same body characteristic corresponding to the sample source of houses picture, N2 >=0;
According to similar point of picture of the sample source of houses picture determining and demand source of houses in each sample source of houses The corresponding recommendation source of houses of picture.
Further, the recommendation information of real estate includes recommending source of houses address and/or recommending source of houses room rate, and the source of houses is searched Rope device further include:
The source of houses obtains module, for being obtained and the recommendation according to the recommendation source of houses address and/or recommendation source of houses room rate The matched similar source of houses of the source of houses, and obtain the similar information of real estate of the similar source of houses;
Second sending module, for the similar information of real estate to be sent to described search terminal.
Further, the source of houses searcher further include:
Address acquisition module, for being referred to according to the navigation when receiving the navigation instruction of described search terminal transmission It enables and determines the target source of houses in the recommendation source of houses, and obtain the destination address of the target source of houses and the end of described search terminal Hold address;
Route Generation module, for generating navigation routine according to the destination address and the terminal address, and will navigation Route is back to described search terminal.
Further, the recommendation source of houses includes more than two sources of houses,
First sending module is also used to obtain the recommendation information of real estate for recommending the source of houses, and is based on predetermined order Rule is ranked up the recommendation information of real estate, and the recommendation information of real estate after sequence is sent to described search terminal;
The source of houses searcher further include:
Rule adjustment module, for obtaining described search terminal to the operation note for recommending information of real estate, and according to The operation note is adjusted the predetermined order rule.
Wherein, the above-mentioned function based on modules in source of houses searcher is realized and above-mentioned source of houses searching method embodiment In each step it is corresponding, function and realization process no longer repeat one by one here.
In addition, the embodiment of the present invention also provides a kind of computer readable storage medium.
Source of houses search program is stored on computer readable storage medium of the present invention, wherein the source of houses search program is located When managing device execution, realize such as the step of above-mentioned source of houses searching method.
Wherein, source of houses search program, which is performed realized method, can refer to each reality of source of houses searching method of the present invention Example is applied, details are not described herein again.
It should be noted that, in this document, the terms "include", "comprise" or its any other variant are intended to non-row His property includes, so that the process, method, article or the system that include a series of elements not only include those elements, and And further include other elements that are not explicitly listed, or further include for this process, method, article or system institute it is intrinsic Element.In the absence of more restrictions, the element limited by sentence "including a ...", it is not excluded that including being somebody's turn to do There is also other identical elements in the process, method of element, article or system.
The serial number of the above embodiments of the invention is only for description, does not represent the advantages or disadvantages of the embodiments.
Through the above description of the embodiments, those skilled in the art can be understood that above-described embodiment side Method can be realized by means of software and necessary general hardware platform, naturally it is also possible to by hardware, but in many cases The former is more preferably embodiment.Based on this understanding, technical solution of the present invention substantially in other words does the prior art The part contributed out can be embodied in the form of software products, which is stored in one as described above In storage medium (such as ROM/RAM, magnetic disk, CD), including some instructions are used so that terminal device (it can be mobile phone, Computer, server, air conditioner or network equipment etc.) execute method described in each embodiment of the present invention.
The above is only a preferred embodiment of the present invention, is not intended to limit the scope of the invention, all to utilize this hair Equivalent structure or equivalent flow shift made by bright specification and accompanying drawing content is applied directly or indirectly in other relevant skills Art field, is included within the scope of the present invention.

Claims (10)

1. a kind of source of houses searching method, which is characterized in that the source of houses searching method includes:
When receiving the source of houses search instruction of search terminal transmission, demand source of houses figure is obtained according to the source of houses search instruction Piece;
Example segmentation is carried out to the demand source of houses picture based on default parted pattern, obtains source of houses main body example and non-source of houses master Body example, and feature extraction is carried out to the source of houses main body example and the non-source of houses main body example respectively, obtain corresponding room Source body feature and non-source of houses body feature;
Carry out source of houses retrieval in default source of houses library according to the source of houses body feature and the non-source of houses body feature, determine with The corresponding recommendation source of houses of the demand source of houses picture;
The recommendation information of real estate for recommending the source of houses is obtained, and the recommendation information of real estate is sent to described search terminal.
2. source of houses searching method as described in claim 1, which is characterized in that described to be based on default parted pattern to the demand Source of houses picture carries out example segmentation, obtains source of houses main body example and the step of non-source of houses main body example includes:
The demand source of houses picture is detected by predeterminable area candidate network RPN, determines the demand source of houses picture Interest region;
Process of convolution is carried out to the demand source of houses picture by default convolutional layer, obtains global feature figure, and according to described emerging It is extracted from the global feature figure and obtains corresponding interest provincial characteristics figure in interesting region;
Pixel correction is carried out to the interest provincial characteristics figure, and by presetting full convolutional network FCN to the region of interest after correction Characteristic of field figure is analyzed, with the example types in the determination interest region;
The demand source of houses picture is split according to the example types in the interest region, obtains source of houses main body example and non- Source of houses main body example.
3. source of houses searching method as described in claim 1, which is characterized in that described according to the source of houses body feature and described Non- source of houses body feature carries out source of houses retrieval in default source of houses library, determines the recommendation source of houses corresponding with the demand source of houses picture The step of include:
The sample source of houses picture of each sample source of houses is obtained in default source of houses library, and obtains the sample master of the sample source of houses picture Body characteristics and non-sample body feature;
The sample body feature and the source of houses body feature are compared, determine the demand source of houses picture and the sample The same body characteristic of this source of houses picture, and the non-sample body feature and the non-source of houses body feature are carried out pair Than determining demand source of houses picture nonbody characteristic identical as the sample source of houses picture;
According to the same body characteristic and the identical nonbody characteristic in each sample source of houses determine with it is described The corresponding recommendation source of houses of demand source of houses picture.
4. source of houses searching method as claimed in claim 3, which is characterized in that described according to the same body characteristic and institute It states identical nonbody characteristic and determines the step for recommending the source of houses corresponding with the demand source of houses picture in each sample source of houses Suddenly include:
The same body characteristic and identical nonbody characteristic are substituting to and preset in similar point of formula, to calculate the sample Similar point of the picture of this source of houses picture, wherein described preset similar point of formula are as follows:
P=k1*N1+k2*N2
Wherein, P is similar point of picture of the sample source of houses picture;
K1 is the first default feature specific gravity, k1 > 0;
K2 is the second default feature specific gravity, k2 > 0;
N1 is same body characteristic corresponding to the sample source of houses picture, N1 >=0;
N2 is same body characteristic corresponding to the sample source of houses picture, N2 >=0;
According to similar point of picture of the sample source of houses picture determining and demand source of houses picture in each sample source of houses The corresponding recommendation source of houses.
5. source of houses searching method as described in claim 1, which is characterized in that the recommendation information of real estate includes with recommending the source of houses Location and/or recommendation source of houses room rate,
It is described to obtain the recommendation information of real estate for recommending the source of houses, and the recommendation information of real estate is sent to described search terminal The step of after, further includes:
According to the recommendation source of houses address and/or recommend the acquisition of source of houses room rate and the matched similar source of houses of the recommendation source of houses, and Obtain the similar information of real estate of the similar source of houses;
The similar information of real estate is sent to described search terminal.
6. source of houses searching method as described in claim 1, which is characterized in that described to obtain the recommendation source of houses for recommending the source of houses Information, and after the step of recommendation information of real estate is sent to described search terminal, further includes:
When receiving the navigation instruction of described search terminal transmission, determined in the recommendation source of houses according to the navigation instruction The target source of houses, and obtain the destination address of the target source of houses and the terminal address of described search terminal;
Navigation routine is generated according to the destination address and the terminal address, and navigation routine is back to described search end End.
7. such as source of houses searching method described in any one of claims 1 to 6, which is characterized in that the recommendation source of houses includes two A above source of houses,
It is described to obtain the recommendation information of real estate for recommending the source of houses, and the recommendation information of real estate is sent to described search terminal The step of include:
The recommendation information of real estate for recommending the source of houses is obtained, and the recommendation information of real estate is arranged based on predetermined order rule Sequence, and the recommendation information of real estate after sequence is sent to described search terminal;
It is described to obtain the recommendation information of real estate for recommending the source of houses, and the recommendation information of real estate is sent to described search terminal The step of after, further includes:
Described search terminal is obtained to the operation note for recommending information of real estate, and according to the operation note to described default Ordering rule is adjusted.
8. a kind of source of houses searcher, which is characterized in that the source of houses searcher includes:
Picture obtains module, for being referred to according to source of houses search when receiving the source of houses search instruction of search terminal transmission Enable acquisition demand source of houses picture;
Characteristic extracting module obtains the source of houses for carrying out example segmentation to the demand source of houses picture based on default parted pattern Main body example and non-source of houses main body example, and feature is carried out to the source of houses main body example and the non-source of houses main body example respectively It extracts, obtains corresponding source of houses body feature and non-source of houses body feature;
Source of houses retrieval module, for according to the source of houses body feature and the non-source of houses body feature in default source of houses library into The retrieval of having sexual intercourse source, determines the recommendation source of houses corresponding with the demand source of houses picture;
First sending module is sent for obtaining the recommendation information of real estate for recommending the source of houses, and by the recommendation information of real estate To described search terminal.
9. a kind of source of houses searches for equipment, which is characterized in that the source of houses search equipment includes processor, memory and storage On the memory and the source of houses search program that can be executed by the processor, wherein the source of houses search program is by the place When managing device and executing, the step of realizing source of houses searching method as described in any one of claims 1 to 7.
10. a kind of computer readable storage medium, which is characterized in that be stored with the source of houses on the computer readable storage medium and search Suo Chengxu, wherein realizing the room as described in any one of claims 1 to 7 when the source of houses search program is executed by processor The step of source searching method.
CN201910202677.6A 2019-03-15 2019-03-15 Source of houses searching method, device, equipment and computer readable storage medium Pending CN110096609A (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN201910202677.6A CN110096609A (en) 2019-03-15 2019-03-15 Source of houses searching method, device, equipment and computer readable storage medium

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN201910202677.6A CN110096609A (en) 2019-03-15 2019-03-15 Source of houses searching method, device, equipment and computer readable storage medium

Publications (1)

Publication Number Publication Date
CN110096609A true CN110096609A (en) 2019-08-06

Family

ID=67443269

Family Applications (1)

Application Number Title Priority Date Filing Date
CN201910202677.6A Pending CN110096609A (en) 2019-03-15 2019-03-15 Source of houses searching method, device, equipment and computer readable storage medium

Country Status (1)

Country Link
CN (1) CN110096609A (en)

Cited By (3)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN110941609A (en) * 2019-10-12 2020-03-31 贝壳技术有限公司 Multi-dimensional searching method and system
CN112487310A (en) * 2020-10-23 2021-03-12 贝壳技术有限公司 Cell recommendation method and system
CN112699289A (en) * 2020-12-30 2021-04-23 上海瑞家信息技术有限公司 House resource information aggregation display method and device, electronic equipment and computer readable medium

Citations (3)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN102521233A (en) * 2010-11-02 2012-06-27 微软公司 Adaptive image retrieval database
CN108416066A (en) * 2018-03-29 2018-08-17 平安好房(上海)电子商务有限公司 Image searching method, user equipment, storage medium and picture searching device
CN109389129A (en) * 2018-09-15 2019-02-26 北京市商汤科技开发有限公司 A kind of image processing method, electronic equipment and storage medium

Patent Citations (3)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN102521233A (en) * 2010-11-02 2012-06-27 微软公司 Adaptive image retrieval database
CN108416066A (en) * 2018-03-29 2018-08-17 平安好房(上海)电子商务有限公司 Image searching method, user equipment, storage medium and picture searching device
CN109389129A (en) * 2018-09-15 2019-02-26 北京市商汤科技开发有限公司 A kind of image processing method, electronic equipment and storage medium

Non-Patent Citations (1)

* Cited by examiner, † Cited by third party
Title
郭耀等: "基于小区域特征的图像检索方法", 《计算机工程》 *

Cited By (5)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN110941609A (en) * 2019-10-12 2020-03-31 贝壳技术有限公司 Multi-dimensional searching method and system
CN110941609B (en) * 2019-10-12 2023-10-20 贝壳找房(北京)科技有限公司 Multi-dimensional searching method and system
CN112487310A (en) * 2020-10-23 2021-03-12 贝壳技术有限公司 Cell recommendation method and system
CN112487310B (en) * 2020-10-23 2023-02-17 贝壳技术有限公司 Cell recommendation method and system
CN112699289A (en) * 2020-12-30 2021-04-23 上海瑞家信息技术有限公司 House resource information aggregation display method and device, electronic equipment and computer readable medium

Similar Documents

Publication Publication Date Title
CN108694223B (en) User portrait database construction method and device
Bennett et al. Inferring and using location metadata to personalize web search
US9020529B2 (en) Computer based location identification using images
CN102591868B (en) System and method for automatic generation of photograph guide
CN110413877A (en) A kind of resource recommendation method, device and electronic equipment
CN107291888B (en) Machine learning statistical model-based living recommendation system method near living hotel
CN107424043A (en) A kind of Products Show method and device, electronic equipment
CN109564571A (en) Utilize the inquiry recommended method and system of search context
CN110069654A (en) Source of houses searching method, device, equipment and computer readable storage medium
CN111061946A (en) Scenario content recommendation method and device, electronic equipment and storage medium
JP5732441B2 (en) Information recommendation method, apparatus and program
CN105512180B (en) A kind of search recommended method and device
CN111191133B (en) Service search processing method, device and equipment
CN104169910A (en) Method to evaluate the geographic popularity of geographically located user-generated content items
CN110083762A (en) Source of houses searching method, device, equipment and computer readable storage medium
CN110096609A (en) Source of houses searching method, device, equipment and computer readable storage medium
CN110019645A (en) Index base construction method, searching method and device
CN110264283B (en) Popularization resource display method and device
CN110263840B (en) Line analysis method, device, program product and storage medium
CN112241489A (en) Information pushing method and device, readable storage medium and computer equipment
CN111159563A (en) Method, device and equipment for determining user interest point information and storage medium
CN112989169A (en) Target object identification method, information recommendation method, device, equipment and medium
CN108897750A (en) Merge the personalized location recommendation method and equipment of polynary contextual information
CN113469752A (en) Content recommendation method and device, storage medium and electronic equipment
CN116762070A (en) System, method, computing platform, and storage medium for providing image recommendations

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

Application publication date: 20190806

WD01 Invention patent application deemed withdrawn after publication