CN105469116B - A kind of child's knowledge figure and data extending method based on human-computer interaction - Google Patents
A kind of child's knowledge figure and data extending method based on human-computer interaction Download PDFInfo
- Publication number
- CN105469116B CN105469116B CN201510862736.4A CN201510862736A CN105469116B CN 105469116 B CN105469116 B CN 105469116B CN 201510862736 A CN201510862736 A CN 201510862736A CN 105469116 B CN105469116 B CN 105469116B
- Authority
- CN
- China
- Prior art keywords
- human
- computer interaction
- subject image
- data
- class data
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Active
Links
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/40—Software arrangements specially adapted for pattern recognition, e.g. user interfaces or toolboxes therefor
- G06F18/41—Interactive pattern learning with a human teacher
Landscapes
- Engineering & Computer Science (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Theoretical Computer Science (AREA)
- Bioinformatics & Computational Biology (AREA)
- Artificial Intelligence (AREA)
- Bioinformatics & Cheminformatics (AREA)
- Life Sciences & Earth Sciences (AREA)
- Data Mining & Analysis (AREA)
- Evolutionary Biology (AREA)
- Evolutionary Computation (AREA)
- Physics & Mathematics (AREA)
- General Engineering & Computer Science (AREA)
- General Physics & Mathematics (AREA)
- Human Computer Interaction (AREA)
- Management, Administration, Business Operations System, And Electronic Commerce (AREA)
Abstract
The invention discloses a kind of, and the child based on human-computer interaction knows figure and data extending method, man-machine interactive platform includes server unit and multiple client unit, the following steps are included: initialization data is stored in server unit, initialization data is A class data for child's knowledge figure and data extending method;Subject image is obtained by shooting or network and is uploaded to server unit, and shooting or the subject image obtained are B class data;For the jobbie image of B class data, the weight of achievement score obtained by each human-computer interaction process is calculated separately according to the result of multiple or multiple human-computer interaction processes, size according to weight determines the final object names of this subject image, and converts A class data from B class data for it.Data untagged can be screened automatically using the intelligent robot of the method for the present invention, marked, and be translated into flag data, with expanding data capacity.
Description
Technical field
The present invention relates to a kind of child of man-machine interactive platform know drawing method, more particularly, to one kind by new data into
Row is automatic to be marked to expand the child's knowledge figure and data extending method of content.
Background technique
In recent years, the technologies such as computer, internet, artificial intelligence achieve breakthrough development, e.g., in hardware device side
Face, high-performance, low cost, low-power consumption computer platform be rapidly developed, the terminal devices such as smart phone are answered extensively
With;In terms of network communication, the universal of Wifi, 3G, 4G allows people whenever and wherever possible by wireless network high speed transmission data;
In terms of speech recognition, the speech recognition technology of the companies such as Google, Microsoft, Baidu exploitation is high to the discrimination of Chinese or English
Up to 95% or more;In terms of image recognition, human testing tends to be mature with face recognition technology and widely applies.This this technology
Development is that the research and development of intelligent Service humanoid robot and application are laid a good foundation.Currently, the application of robot is just from past profession
Change field by the normal sphere of life of mesh for expanding to people greatly, especially can provide accompany and attend to, the intelligent Service type machine of the services such as early education
Device people is just becoming using hot spot.
But existing service humanoid robot still has very big room for promotion in intelligence degree, service content.
For example, needing the sample database of a large amount of tape labels for training in terms of the image recognition to different objects, and acquire these
The workload of data itself is very huge, needs to take a substantial amount of time and manpower, is far from the work that a few peoples can complete
Make, this is to influence robot recognition capability, the major reason of level of intelligence.
Summary of the invention
The child that the object of the present invention is to provide a kind of based on human-computer interaction knows figure and data extending method, have intelligence,
Automatically, practical advantage can automatically screen data untagged using the intelligent robot of the method for the present invention, be marked, and
It is translated into flag data, with expanding data capacity, the intelligent level of hoisting machine people.
Need to acquire a large amount of bands in terms of the image recognition to different objects to solve service humanoid robot in the prior art
The technical issues of sample data of label is used to train, time-consuming and laborious and influence robot recognition capability and level of intelligence, this hair
Bright to provide a kind of child's knowledge figure and data extending method based on human-computer interaction, wherein man-machine interactive platform includes server list
Member and multiple client unit, server unit and client unit are communicated by wired or wireless way, and the child knows figure
And data extending method the following steps are included:
One, by the database of initialization data deposit server unit, initialization data includes subject image and correspondence
Object names;Initialization data is divided into A class data and distributes unique number for each subject image of A class data,
The number is for identifying the subject image;
Two, subject image is shot by client unit or obtains subject image from network, obtained by shooting or from network
Subject image be uploaded in the database of server unit;B class number is divided by shooting or from the subject image that network obtains
Unique number, described number subject image for identification are distributed according to and for each subject image of B class data;
Three, human-computer interaction participant is loaded into A class data or B in server unit database by client unit at random
The subject image of class data, and corresponding object names are answered according to the subject image of loading;
Four, the subject image number of A class data is loaded into for a human-computer interaction process, according to returning for human-computer interaction participant
It answers correctness and calculates gained achievement score, and record human-computer interaction participant and returned for each subject image of B class data
The object names answered;
Five, the achievement score obtained step 4 and human-computer interaction participant are directed to each subject image of B class data
The object names of answer are uploaded to server unit together, the human-computer interaction result returned as the secondary human-computer interaction process;
Six, according to the mode of step 1 to step 5, by all knots for participating in human-computer interaction process by client unit
Fruit is uploaded to server unit;
Seven, it for the jobbie image of B class data, is calculated separately according to the result of multiple or multiple human-computer interaction processes
The weight of achievement score obtained by each human-computer interaction process, and the size according to weight determines the final object of this subject image
Title;
Eight, the final object names of certain the B class data subject image obtained according to step 7, to this B class data
Subject image is marked, and converts A class data from B class data for it.
Further, a kind of child based on human-computer interaction of the present invention knows figure and data extending method, wherein in above-mentioned step
In rapid four, the subject image number for being directed to a human-computer interaction process and being loaded into A class data, according to returning for human-computer interaction participant
It answers correctness and calculates gained achievement score, specifically includes the following steps:
(1) by accumulation calculating, the secondary human-computer interaction process is loaded into the subject image sum of A class data, and is recorded as m;
(2) number according to the subject image for being loaded into every A class data, the object name that human-computer interaction participant is answered
Claim to compare with the object names of this subject image corresponding storage in the database, be judged as if unanimously and answer correctly,
It is judged as erroneous answers if inconsistent, statistics answers correct number, and is recorded as m1
(3) with answering correct number divided by the total number of subject image for being loaded into A class data, using gained divisor as at
Achievement score, and it is denoted as s=m1/m。
Further, a kind of child based on human-computer interaction of the present invention knows figure and data extending method, wherein in above-mentioned step
In rapid seven, the basis is multiple or the result of multiple human-computer interaction processes calculates separately achievement obtained by each human-computer interaction process point
Several weights, and the size according to weight determines the final object names of this subject image, specifically includes the following steps:
(1) it is directed to a certain subject image of B class data, the number for returning to human-computer interaction result is set as n, by n times people
The achievement score that machine interactive process returns is set as set si={ s1, s2, s3..., sn};The people that n times human-computer interaction process is returned
The object names that machine interaction participant answers for this subject image are set as set di={ d1, d2, d3..., dn};
(2) the object figure is directed to according to the whether identical human-computer interaction participant returned to n times human-computer interaction process of content
The object names that picture is answered distinguish, and only retain primary mode according to identical content and obtain the different object names of k kind, remember
For set dj={ d1, d2, d3..., dk}.
(3) according to set dj={ d1, d2, d3..., dkIn each element, with dj=diFor condition accumulation calculating each element
Corresponding achievement score, as the corresponding score weight w of each elementi,
wi=∑ (si|dj=di)
(4) to the corresponding score weight w of each elementiIt is normalized, as the corresponding normalization weight of each element
(5) for step (4) obtain as a result, the corresponding normalization weight of each element is sorted from large to small and successively
It is cumulative, until being greater than the threshold value t of setting until accumulative normalization weights sum has been added, wherein 0 < t < 1;
(6) judged according to the number that accumulative normalization weight is added in step (5),
If the number j=1 that accumulative normalization weight is added: if directly by the corresponding element d of maximum normalization weightj
As the final object names to the subject image;
If the number j=2 of accumulative normalization weight is added, 3: being judged by the method that the meaning of a word is analysed and compared corresponding
Several element djWhether it is synonym, if synonym or is the word comprising justice, then simultaneously by these elements djAs the object
The final object names of body image;
If the number j=2 of accumulative normalization weight is added, 3 when corresponding several element djIt is not synonym, or is added
When the number j > 3 of accumulative normalization weight, then it represents that condition is unsatisfactory for not determining final object names, and is returned by increasing
The number of the Huis' machine interaction results continues to execute step (1) to step (5), until determining the final object of the subject image
Title.
Further, a kind of child based on human-computer interaction of the present invention knows figure and data extending method, wherein the threshold value
The value of t is 0.7.
Further, a kind of child based on human-computer interaction of the present invention knows figure and data extending method, wherein the A class
The corresponding one or more object names of a subject image in data, and an object names correspond to multiple subject images.
Further, a kind of child based on human-computer interaction of the present invention knows figure and data extending method, wherein the A class
The subject image of data and B class data is all made of the image of fixed size, and object is made to be located at image middle position, and with jpg,
The storage of bmp format.
Further, a kind of child based on human-computer interaction of the present invention knows figure and data extending method, wherein the A class
The object names of data are indicated using one or more Chinese-character words.
A kind of child based on human-computer interaction of the present invention knows figure and data extending method compared with prior art, has following
Advantage: the invention proposes a kind of human-computer interactions to know figure and online data extending method, in practical application, allow server unit and
Robot as client unit is communicated by network, can while making one to carry out human-computer interaction knowledge figure with robot
To be screened, be marked to the new data of acquisition, with automatic expanding data appearance;The a large amount of new datas obtained can be also used for passing through
Machine learning training, the level of intelligence of hoisting machine people.This mode does not need to acquire a large amount of flag data in advance, saves
Time and manpower, and improve the recognition capability and level of intelligence of robot, there is very strong practicability.
The illustrated embodiment child based on human-computer interaction a kind of to the present invention knows figure and data extending side with reference to the accompanying drawing
Method is described in further detail:
Detailed description of the invention
Fig. 1 is the generation stream that a kind of child based on human-computer interaction of the present invention knows A class data in figure and data extending method
Cheng Tu;
Fig. 2 is that a kind of child based on human-computer interaction of the present invention knows the frame that B class data are obtained in figure and data extending method
Line chart.
Fig. 3 is that human-computer interaction participant is logical in a kind of child's knowledge figure and data extending method based on human-computer interaction of the present invention
Cross the wire figure that client unit is loaded into A class data or B class data at random;
Fig. 4 is that human-computer interaction participant is logical in a kind of child's knowledge figure and data extending method based on human-computer interaction of the present invention
Cross the flow chart that client unit carries out human-computer interaction;
Fig. 5 is to convert A for B class data in a kind of child's knowledge figure and data extending method based on human-computer interaction of the present invention
The wire figure of class data.
Specific embodiment
Firstly the need of explanation, in the present patent application file, the child and human-computer interaction participant should make identical general
It reads and understands, refer both to the personnel for participating in human-computer interaction by client unit, the robot refers to client unit.The present invention one
Child of the kind based on human-computer interaction knows figure and data extending method, man-machine interactive platform include server unit and multiple client
Unit, server unit and client unit are communicated by wired or wireless way, and child's knowledge figure and data extending method include
Following steps:
One, by the database of initialization data deposit server unit, initialization data includes subject image and correspondence
Object names;Initialization data is divided into A class data and distributes unique number for each subject image of A class data,
The number is for identifying the subject image;
Two, subject image is shot by client unit or obtains subject image from network, obtained by shooting or from network
Subject image be uploaded in the database of server unit;B class number is divided by shooting or from the subject image that network obtains
Unique number, described number subject image for identification are distributed according to and for each subject image of B class data;
Three, human-computer interaction participant is loaded into A class data or B in server unit database by client unit at random
The subject image of class data, and corresponding object names are answered according to the subject image of loading;
Four, the subject image number of A class data is loaded into for a human-computer interaction process, according to returning for human-computer interaction participant
It answers correctness and calculates gained achievement score, and record human-computer interaction participant and returned for each subject image of B class data
The object names answered;
Five, the achievement score obtained step 4 and human-computer interaction participant are directed to each subject image of B class data
The object names of answer are uploaded to server unit together, the human-computer interaction result returned as the secondary human-computer interaction process;
Six, according to the mode of step 1 to step 5, by all knots for participating in human-computer interaction process by client unit
Fruit is uploaded to server unit;
Seven, it for the jobbie image of B class data, is calculated separately according to the result of multiple or multiple human-computer interaction processes
The weight of achievement score obtained by each human-computer interaction process, and the size according to weight determines the final object of this subject image
Title;
Eight, the final object names of certain the B class data subject image obtained according to step 7, to this B class data
Subject image is marked, and converts A class data from B class data for it.
In above-mentioned steps four, the subject image number for being directed to a human-computer interaction process and being loaded into A class data, according to people
The answer correctness of machine interaction participant calculates gained achievement score, specifically includes the following steps:
(1) by accumulation calculating, the secondary human-computer interaction process is loaded into the subject image sum of A class data, and is recorded as m;
(2) number according to the subject image for being loaded into every A class data, the object name that human-computer interaction participant is answered
Claim to compare with the object names of this subject image corresponding storage in the database, be judged as if unanimously and answer correctly,
It is judged as erroneous answers if inconsistent, statistics answers correct number, and is recorded as m1
(3) with answering correct number divided by the total number of subject image for being loaded into A class data, using gained divisor as at
Achievement score, and it is denoted as s=m1/m。
In above-mentioned steps seven, the basis is multiple or the result of multiple human-computer interaction processes calculates separately each man-machine friendship
The weight of achievement score obtained by mutual process, and the size according to weight determines the final object names of this subject image, specifically
The following steps are included:
(1) it is directed to a certain subject image of B class data, the number for returning to human-computer interaction result is set as n, by n times people
The achievement score that machine interactive process returns is set as set si={ s1, s2, s3..., sn};The people that n times human-computer interaction process is returned
The object names that machine interaction participant answers for this subject image are set as set di={ d1, d2, d3..., dn};
(2) the object figure is directed to according to the whether identical human-computer interaction participant returned to n times human-computer interaction process of content
The object names that picture is answered distinguish, and only retain primary mode according to identical content and obtain the different object names of k kind, remember
For set dj={ d1, d2, d3..., dk}.
(3) according to set dj={ d1, d2, d3..., dkIn each element, with dj=diFor condition accumulation calculating each element
Corresponding achievement score, as the corresponding score weight w of each elementi,
wi=∑ (si|dj=di)
(4) to the corresponding score weight w of each elementiIt is normalized, as the corresponding normalization weight of each element
(5) for step (4) obtain as a result, the corresponding normalization weight of each element is sorted from large to small and successively
It is cumulative, until being greater than the threshold value t of setting until accumulative normalization weights sum has been added, wherein 0 < t < 1;
(6) judged according to the number that accumulative normalization weight is added in step (5),
If the number j=1 that accumulative normalization weight is added: if directly by the corresponding element d of maximum normalization weightj
As the final object names to the subject image;
If the number j=2 of accumulative normalization weight is added, 3: being judged by the method that the meaning of a word is analysed and compared corresponding
Several element djWhether it is synonym, if synonym or is the word comprising justice, then simultaneously by these elements djAs the object
The final object names of body image;
If the number j=2 of accumulative normalization weight is added, 3 when corresponding several element djIt is not synonym, or is added
When the number j > 3 of accumulative normalization weight, then it represents that condition is unsatisfactory for not determining final object names, and is returned by increasing
The number of the Huis' machine interaction results continues to execute step (1) to step (5), until determining the final object of the subject image
Title.
As specific embodiment, the value of the threshold value t can be set as 0.7.It should be noted that the value of threshold value t can root
It is adjusted according to required precision, the final object names of the bigger determination of value are more accurate, and practical application shows that t value 0.7 can be realized
The object of the invention.In practical application, the corresponding one or more object names of a subject image of A class data are allowed, and allow one
Object names correspond to multiple subject images;The subject image of A class data and B class data is all made of the image of fixed size, and makes
Object is located at image middle position, and with the storage of jpg, bmp format;One or more Chinese can be used in the object names of A class data
Words language indicates.
To help skilled in the art to understand the present invention, one kind of the present invention is based in a manner of specific embodiment below
The child's knowledge figure and data extending method of human-computer interaction are further described.
Method of the invention is directed to child's early education, is carried out with the man-machine interaction mode of looking-and-understanding.Man-machine interactive platform packet
Include server unit and multiple client unit;Child knows figure and data extending method by data initialization, human-computer interaction, data
Handle three process compositions, detailed process are as follows:
In using the preceding database by the storage of A class data to server unit, by B class data in using preceding or use
It is also stored into the database of server unit.
A class data include subject image+object names, i.e. marked sample;
B class data only have subject image, i.e. unmarked sample;
Wherein, subject image using background simple, fixed size image, and object is allowed to be located at the middle position of image,
With the storage of the image formats such as jpg, Bmp;Object names are indicated using one or more Chinese-character words.
For A class data, classifies by object names and each object names descriptor is made to correspond to multiple different objects
Body image all distributes unique number for it and it is made to correspond to one or more different object name to each subject image
Claim.As shown in Figure 1, the generation standard of A class data can carry out as follows:
I. start to input;
Ii. shooting is free of the background image I of object0, simple background also can be directly used;
Iii. I is shot0It include the image I of object under background1;
Iv. subject image I=I1-I0;
V. speech recognition or text mode input object names, form one or several words and describe as object names
Symbol.
For B class data, as shown in Fig. 2, image directly can be obtained from internet, it can also be by the machine as client
Device people shoots image in daily activities, and is uploaded to server, by the subject image point that server is each B class data
With unique number.
As shown in Figure 3 and Figure 4, in human-computer interaction process, A class data and B that the robot as client unit is read
Class data are all from the database of server unit, and detailed process is as follows:
(1) start human-computer interaction process, i.e. knowledge figure process, robot is loaded into data at random and shows image, puts question to that " this is
What ", loaded data are also possible to B class data either A class data;
(2) human-computer interaction participant answers object names by voice mode, and voice is known by speech recognition by robot
Not at text, as answer;
(3) be A class data if what is be loaded into: if answer and object names compared, provide correct/error and adjudicate;
(4) if be loaded into be B class data: if directly answer will be recorded;
(5) in a human-computer interaction process, the subject image number scale for being loaded onto A class data is m, identifies correct number note
For m1, then the achievement score of this human-computer interaction process are as follows: s=100 × m1/m。
After human-computer interaction process, by the identification gained achievement score s for A class data subject image, and
To the correspondence answer upload of each subject image of B class data to server unit.
As shown in figure 5, the data that the database of server unit is used comprising all clients unit (robot), more
Robot is connected to the same server unit by mobile Internet, by the achievement score of human-computer interaction each time and to B
The answer upload of class data allows server unit to be distinguished according to the result of multiple or multiple human-computer interaction processes to server unit
The weight of achievement score obtained by each human-computer interaction process is calculated, and the size according to weight determines the accurate of this subject image
Answer, as final object names, the specific method is as follows:
For a certain subject image of B class data, the number for returning to human-computer interaction result is set as n, by the man-machine friendship of n times
The score that mutual process returns is set as set si, si={ s1, s2, s3..., sn};The human-computer interaction that n times human-computer interaction process is returned
The object names that participant answers for the subject image are set as set di, di={ d1, d2, d3..., dn}.
It is returned according to the whether identical human-computer interaction participant returned to n times human-computer interaction process of content for the subject image
The object names answered distinguish, and only retain primary mode according to identical content and obtain the different object names of k kind, are denoted as collection
Close dj={ d1, d2, d3..., dk}.
According to set dj={ d1, d2, d3..., dkIn each element, with dj=diIt is corresponding for condition accumulation calculating each element
Achievement score be denoted as w as the corresponding score weight of each elementi。
wi=∑ (si|dj=di)
Score weight w corresponding to each elementiIt is normalized, as the corresponding normalization weight of each element, note
For
The corresponding normalization weight of each element is sorted from large to small and is successively added up, until accumulative normalization has been added
Until weights sum is greater than the threshold value t of setting, wherein 0 < t < 1;T value is bigger, and the final object names determined are more accurate, real
Border application shows that the object of the invention can be realized in t value 0.7.
Judged according to the number that accumulative normalization weight is added, if the number j of accumulative normalization weight is added
=1: then directly by the corresponding element d of maximum normalization weightjAs the final object names to the subject image;
If the number j=2 of accumulative normalization weight is added, 3: being judged by the method that the meaning of a word is analysed and compared corresponding
Several element djWhether it is synonym, if synonym or is the word (such as automobile, car) comprising justice, then simultaneously by these
Element djFinal object names as the subject image;
If the number j=2 of accumulative normalization weight is added, 3 when corresponding several element djIt is not synonym (such as vapour
Vehicle, aircraft), or when the number j > 3 of accumulative normalization weight is added, then it represents that condition is unsatisfactory for not determining final object name
Claim, and return to the number of human-computer interaction result by increasing, continues according to the above method until determining the final of the subject image
Object names.
According to determining final object names, the subject image of this B class data is marked, by it from B class number
According to being converted into A class data.
Above embodiments are only the descriptions carried out to the preferred embodiment of the present invention, and model not is claimed to the present invention
The restriction for enclosing progress, under the premise of not departing from design principle of the present invention and spirit, this field engineers and technicians are according to this hair
The various forms of deformations that bright technical solution is made, should all fall into protection scope determined by claims of the present invention.
Claims (7)
1. a kind of child based on human-computer interaction knows figure and data extending method, man-machine interactive platform used in this method includes
Server unit and multiple client unit, server unit and client unit are communicated by wired or wireless way, special
Sign exists: the child know figure and data extending method the following steps are included:
One, by the database of initialization data deposit server unit, initialization data includes subject image and corresponding object
Body title;Initialization data is divided into A class data and distributes unique number for each subject image of A class data, it is described
Number is for identifying the subject image;
Two, subject image is shot by client unit or obtains subject image from network, by shooting or the object obtained from network
Body image is uploaded in the database of server unit;B class data are divided into simultaneously by shooting or from the subject image that network obtains
Unique number is distributed for each subject image of B class data, it is described to number the subject image for identification;
Three, human-computer interaction participant is loaded into A class data or B class number in server unit database by client unit at random
According to subject image, and answer corresponding object names according to the subject image of loading;
Four, the subject image number of A class data is loaded into for a human-computer interaction process, just according to the answer of human-computer interaction participant
Gained achievement score is calculated whether really, and records what human-computer interaction participant answered for each subject image of B class data
Object names;
Five, the achievement score and human-computer interaction participant obtained step 4 is answered for each subject image of B class data
Object names be uploaded to server unit together, the human-computer interaction result returned as the secondary human-computer interaction process;
It six, will be in all results for participating in human-computer interaction process by client unit according to the mode of step 1 to step 5
Reach server unit;
Seven, it for the jobbie image of B class data, is calculated separately every time according to the result of multiple or multiple human-computer interaction processes
The weight of achievement score obtained by human-computer interaction process, and the size according to weight determines the final object name of this subject image
Claim;
Eight, the final object names of certain the B class data subject image obtained according to step 7, to the object of this B class data
Image is marked, and converts A class data from B class data for it.
2. a kind of child based on human-computer interaction described in accordance with the claim 1 knows figure and data extending method, it is characterised in that:
In above-mentioned steps four, the subject image number for being directed to a human-computer interaction process and being loaded into A class data is joined according to human-computer interaction
Gained achievement score is calculated with the answer correctness of person, specifically includes the following steps:
(1) by accumulation calculating, the secondary human-computer interaction process is loaded into the subject image sum of A class data, and is recorded as m;
(2) according to be loaded into every A class data subject image number, the object names that human-computer interaction participant is answered with
The corresponding object names stored compare this subject image in the database, are judged as that answer is correct if consistent, if
It is inconsistent, it is judged as erroneous answers, statistics answers correct number, and is recorded as m1;
(3) with correct number is answered divided by the total number of subject image for being loaded into A class data, divide gained divisor as achievement
Number, and it is denoted as s=m1/m。
3. a kind of child based on human-computer interaction knows figure and data extending method according to claim 2, it is characterised in that:
In above-mentioned steps seven, the basis is multiple or the result of multiple human-computer interaction processes calculates separately each human-computer interaction process institute
The weight of achievement score is obtained, and the size according to weight determines the final object names of this subject image, specifically includes following
Step:
(1) it is directed to a certain subject image of B class data, the number for returning to human-computer interaction result is set as n, by the man-machine friendship of n times
The achievement score that mutual process returns is set as set si={ s1, s2, s3..., sn};The man-machine friendship that n times human-computer interaction process is returned
The object names that mutual participant answers for this subject image are set as set di={ d1, d2, d3..., dn};
(2) it is returned according to the whether identical human-computer interaction participant returned to n times human-computer interaction process of content for the subject image
The object names answered distinguish, and only retain primary mode according to identical content and obtain the different object names of k kind, are denoted as collection
Close dj={ d1, d2, d3..., dk};
(3) according to set dj={ d1, d2, d3..., dkIn each element, with dj=diIt is corresponding for condition accumulation calculating each element
Achievement score, as the corresponding score weight w of each elementi, wi=∑ (si|dj=di);
(4) to the corresponding score weight w of each elementiIt is normalized, as the corresponding normalization weight of each element
(5) for step (4) acquisition as a result, the corresponding normalization weight of each element is sorted from large to small and is successively tired out
Add, until being greater than the threshold value t of setting until accumulative normalization weights sum has been added, wherein 0 < t < 1;
(6) judged according to the number that accumulative normalization weight is added in step (5), if accumulative normalization power is added
The number j=1 of value: then directly by the corresponding element d of maximum normalization weightjAs the final object name to the subject image
Claim;If the number j=2 of accumulative normalization weight is added, 3: being judged by the method that the meaning of a word is analysed and compared corresponding several
Element djWhether it is synonym, if synonym or is the word comprising justice, then simultaneously by these elements djAs the object figure
The final object names of picture;If the number j=2 of accumulative normalization weight is added, 3 when corresponding several element djIt is not synonymous
Word, or when the number j > 3 of accumulative normalization weight is added, then it represents that condition is unsatisfactory for not determining final object names, and
The number that human-computer interaction result is returned by increasing continues to execute step (1) to step (5), until determining the subject image
Final object names.
4. a kind of child based on human-computer interaction described in accordance with the claim 3 knows figure and data extending method, it is characterised in that:
The value of the threshold value t is 0.7.
5. a kind of child based on human-computer interaction described in accordance with the claim 1 knows figure and data extending method, it is characterised in that:
The corresponding one or more object names of a subject image in the A class data, and an object names correspond to multiple object figures
Picture.
6. a kind of child based on human-computer interaction described in accordance with the claim 1 knows figure and data extending method, it is characterised in that:
The subject image of the A class data and B class data is all made of the image of fixed size, and object is made to be located at image middle position,
And with the storage of jpg, bmp format.
7. a kind of child based on human-computer interaction described in accordance with the claim 1 knows figure and data extending method, it is characterised in that:
The object names of the A class data are indicated using one or more Chinese-character words.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201510862736.4A CN105469116B (en) | 2015-12-01 | 2015-12-01 | A kind of child's knowledge figure and data extending method based on human-computer interaction |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201510862736.4A CN105469116B (en) | 2015-12-01 | 2015-12-01 | A kind of child's knowledge figure and data extending method based on human-computer interaction |
Publications (2)
Publication Number | Publication Date |
---|---|
CN105469116A CN105469116A (en) | 2016-04-06 |
CN105469116B true CN105469116B (en) | 2019-01-04 |
Family
ID=55606787
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN201510862736.4A Active CN105469116B (en) | 2015-12-01 | 2015-12-01 | A kind of child's knowledge figure and data extending method based on human-computer interaction |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN105469116B (en) |
Families Citing this family (3)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN106873893B (en) * | 2017-02-13 | 2021-01-22 | 北京光年无限科技有限公司 | Multi-modal interaction method and device for intelligent robot |
CN107589963B (en) * | 2017-09-26 | 2019-05-17 | 维沃移动通信有限公司 | A kind of image processing method, mobile terminal and computer readable storage medium |
CN112464894B (en) * | 2020-12-14 | 2023-09-01 | 深圳市优必选科技股份有限公司 | Interaction method and device and computer equipment |
Citations (4)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN101093572A (en) * | 2007-07-20 | 2007-12-26 | 中国建设银行股份有限公司 | A preposition system and a centralized data processing system |
CN102159134A (en) * | 2008-07-08 | 2011-08-17 | 普罗秋斯生物医学公司 | Ingestible event marker data framework |
CN103035135A (en) * | 2012-11-27 | 2013-04-10 | 北京航空航天大学 | Children cognitive system based on augment reality technology and cognitive method |
CN104157168A (en) * | 2014-08-21 | 2014-11-19 | 鲜于庆 | Intelligent database application system applied to artificial intelligent robot and interaction method |
Family Cites Families (1)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
US7765231B2 (en) * | 2005-04-08 | 2010-07-27 | Rathus Spencer A | System and method for accessing electronic data via an image search engine |
-
2015
- 2015-12-01 CN CN201510862736.4A patent/CN105469116B/en active Active
Patent Citations (4)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN101093572A (en) * | 2007-07-20 | 2007-12-26 | 中国建设银行股份有限公司 | A preposition system and a centralized data processing system |
CN102159134A (en) * | 2008-07-08 | 2011-08-17 | 普罗秋斯生物医学公司 | Ingestible event marker data framework |
CN103035135A (en) * | 2012-11-27 | 2013-04-10 | 北京航空航天大学 | Children cognitive system based on augment reality technology and cognitive method |
CN104157168A (en) * | 2014-08-21 | 2014-11-19 | 鲜于庆 | Intelligent database application system applied to artificial intelligent robot and interaction method |
Also Published As
Publication number | Publication date |
---|---|
CN105469116A (en) | 2016-04-06 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN108536681B (en) | Intelligent question-answering method, device, equipment and storage medium based on emotion analysis | |
EP3989166A1 (en) | Artificial intelligence-based image region recognition method and apparatus, and model training method and apparatus | |
US11238310B2 (en) | Training data acquisition method and device, server and storage medium | |
CN104463101B (en) | Answer recognition methods and system for character property examination question | |
CN107679447A (en) | Facial characteristics point detecting method, device and storage medium | |
CN108491486B (en) | Method, device, terminal equipment and storage medium for simulating patient inquiry dialogue | |
CN107680019A (en) | A kind of implementation method of Examination Scheme, device, equipment and storage medium | |
CN107562918A (en) | A kind of mathematical problem knowledge point discovery and batch label acquisition method | |
CN106295591A (en) | Gender identification method based on facial image and device | |
CN109902912B (en) | Personalized image aesthetic evaluation method based on character features | |
CN110321409B (en) | Auxiliary interview method, device, equipment and storage medium based on artificial intelligence | |
CN103745423B (en) | A kind of shape of the mouth as one speaks teaching system and teaching method | |
CN109902285A (en) | Corpus classification method, device, computer equipment and storage medium | |
CN109783624A (en) | Answer generation method, device and the intelligent conversational system in knowledge based library | |
CN110378736B (en) | Method for evaluating experience satisfaction degree of tourists on natural resources through facial expression recognition | |
CN109800309A (en) | Classroom Discourse genre classification methods and device | |
CN105469116B (en) | A kind of child's knowledge figure and data extending method based on human-computer interaction | |
CN113408570A (en) | Image category identification method and device based on model distillation, storage medium and terminal | |
CN106227836B (en) | Unsupervised joint visual concept learning system and unsupervised joint visual concept learning method based on images and characters | |
CN105224920A (en) | A kind ofly will proceed to the learning test diagnostic system on line under line | |
CN109918641A (en) | Article theme ingredient breakdown method, apparatus, equipment and storage medium | |
CN103473275A (en) | Automatic image labeling method and automatic image labeling system by means of multi-feature fusion | |
CN109284700A (en) | The method of multiple Face datections, storage medium, equipment and system in image | |
CN114639152A (en) | Multi-modal voice interaction method, device, equipment and medium based on face recognition | |
CN107784482A (en) | Recruitment methods, electronic installation and readable storage medium storing program for executing |
Legal Events
Date | Code | Title | Description |
---|---|---|---|
C06 | Publication | ||
PB01 | Publication | ||
C10 | Entry into substantive examination | ||
SE01 | Entry into force of request for substantive examination | ||
GR01 | Patent grant | ||
GR01 | Patent grant |