CN110991322A - Emotion recognition model updating method and device, electronic equipment and medium - Google Patents
Emotion recognition model updating method and device, electronic equipment and medium Download PDFInfo
- Publication number
- CN110991322A CN110991322A CN201911202062.XA CN201911202062A CN110991322A CN 110991322 A CN110991322 A CN 110991322A CN 201911202062 A CN201911202062 A CN 201911202062A CN 110991322 A CN110991322 A CN 110991322A
- Authority
- CN
- China
- Prior art keywords
- recognition model
- emotion recognition
- current image
- server
- current
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Pending
Links
Images
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V40/00—Recognition of biometric, human-related or animal-related patterns in image or video data
- G06V40/10—Human or animal bodies, e.g. vehicle occupants or pedestrians; Body parts, e.g. hands
- G06V40/16—Human faces, e.g. facial parts, sketches or expressions
- G06V40/174—Facial expression recognition
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/21—Design or setup of recognition systems or techniques; Extraction of features in feature space; Blind source separation
- G06F18/214—Generating training patterns; Bootstrap methods, e.g. bagging or boosting
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V40/00—Recognition of biometric, human-related or animal-related patterns in image or video data
- G06V40/10—Human or animal bodies, e.g. vehicle occupants or pedestrians; Body parts, e.g. hands
- G06V40/16—Human faces, e.g. facial parts, sketches or expressions
- G06V40/168—Feature extraction; Face representation
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V40/00—Recognition of biometric, human-related or animal-related patterns in image or video data
- G06V40/10—Human or animal bodies, e.g. vehicle occupants or pedestrians; Body parts, e.g. hands
- G06V40/16—Human faces, e.g. facial parts, sketches or expressions
- G06V40/172—Classification, e.g. identification
Abstract
The invention provides an updating processing method and device of an emotion recognition model, electronic equipment and a storage medium, wherein the updating processing method at a terminal side comprises the following steps: receiving a current emotion recognition model sent by a server after the emotion recognition model of the server is updated each time; when a current image is acquired, performing emotion recognition on the current image by using the current emotion recognition model so as to determine target emotion information corresponding to the current image in a plurality of candidate emotion information; if the current emotion recognition model cannot determine the only target emotion information for the current image, uploading the current image or the characteristic information of the current image to the server, so that the server can determine a training material according to the current image or the characteristic information of the current image, and training the emotion recognition model of the server by using the training material to update the emotion recognition model of the server.
Description
Technical Field
The invention relates to the field of emotion recognition, in particular to an emotion recognition model updating method and device, electronic equipment and a medium.
Background
The terminal may be understood as being configured with a processor and a memory, and further, the processor reads a code in the memory to execute a corresponding function, and may specifically be, for example, a mobile phone, a vehicle-mounted terminal, an image capturing terminal, a tablet computer, and the like. In the prior art, an emotion recognition model can be configured in a terminal, and then the emotion recognition model is used for carrying out emotion recognition on the collected image.
In the prior art, the emotion recognition model is received from the server, and the terminal side cannot correct the emotion recognition model, which is not beneficial to the emotion recognition model to improve the recognition accuracy.
Disclosure of Invention
The invention provides an updating processing method and device of an emotion recognition model, electronic equipment and a medium, and aims to solve the problem that the emotion recognition model cannot be corrected at a terminal side, which is not beneficial to improving the recognition accuracy of the emotion recognition model.
According to a first aspect of the present invention, there is provided a terminal side update processing method of an emotion recognition model, applied to a terminal, including:
receiving a current emotion recognition model sent by a server after the emotion recognition model of the server is updated each time;
when a current image is acquired, performing emotion recognition on the current image by using the current emotion recognition model so as to determine target emotion information corresponding to the current image in a plurality of candidate emotion information;
if the current emotion recognition model cannot determine the only target emotion information for the current image, uploading the current image or the characteristic information of the current image to the server, so that the server can determine a training material according to the current image or the characteristic information of the current image, and training the emotion recognition model of the server by using the training material to update the emotion recognition model of the server.
Optionally, performing emotion recognition on the current image by using the current emotion recognition model to determine target emotion information corresponding to the current image in a plurality of candidate emotion information, where the determining includes:
comparing the current image with a standard image corresponding to each candidate emotion information through the current emotion recognition model, or comparing the feature information in the current image with the feature information of each standard image, and determining the matching evaluation data of the current image and each standard image; the matching evaluation data is used for representing the matching degree of the characteristic information in the current image and the characteristic information of the corresponding standard image by using a numerical value;
and acquiring the target emotion information determined by the current emotion recognition model according to the matching evaluation data.
Optionally, before uploading the current image or the feature information thereof to the server, the method further includes:
and determining that the matching evaluation data of the two standard images with the highest matching degree are the same or the difference value is smaller than a threshold value, so that the current emotion recognition model cannot determine unique target emotion information for the current image.
Optionally, the terminal is a vehicle-mounted terminal, and the current image is an image of a driver and/or a passenger acquired by an in-vehicle image acquisition component.
According to a second aspect of the present invention, there is provided a terminal-side update processing apparatus of an emotion recognition model, including:
the terminal receiving module is used for receiving the current emotion recognition model sent by the server each time the emotion recognition model of the server is updated;
the identification module is used for carrying out emotion identification on the current image by using the current emotion identification model when the current image is obtained so as to determine target emotion information corresponding to the current image in a plurality of candidate emotion information;
and the uploading module is used for uploading the current image or the characteristic information thereof to the server if the current emotion recognition model cannot determine the only target emotion information aiming at the current image, so that the server can determine the current image and/or other images similar to the current image as training materials, train the emotion recognition model in the server and update the emotion recognition model.
According to a third aspect of the present invention, there is provided a server-side update processing method of an emotion recognition model, applied to a server, including:
receiving a current image or characteristic information thereof sent by any one target terminal in a plurality of terminals, wherein the current image is an image of which the current emotion recognition model cannot determine unique target emotion information when the terminal performs emotion recognition on the current image by using the current emotion recognition model; the current emotion recognition model used by the terminal is sent by the server;
determining a training material according to the current image and/or the characteristic information thereof; the training material is at least one of the following: the current image, the feature information of the current image, other similar images similar to the current image and the feature information of the other similar images;
after the calibration information of the training material is obtained, training an emotion recognition model of the server by using the training material and the calibration information so as to update the emotion recognition model of the server;
and sending the emotion recognition model of the server to the plurality of terminals as a current emotion recognition model.
Optionally, the terminal is a vehicle-mounted terminal, and the current image is an image of a driver and/or a passenger acquired by an in-vehicle image acquisition component.
According to a fourth aspect of the present invention, there is provided a server-side update processing apparatus of an emotion recognition model, including:
the server receiving module is used for receiving a current image or characteristic information thereof sent by any one target terminal in a plurality of terminals, wherein the current image is an image of which the current emotion recognition model cannot determine unique target emotion information when the terminal performs emotion recognition on the current image by using the current emotion recognition model; the current emotion recognition model used by the terminal is sent by the server;
the material determining module is used for determining training materials according to the current image and/or the characteristic information thereof; the training material is at least one of the following: the current image, the feature information of the current image, and the feature information of other similar images similar to the current image and the other similar images
The training module is used for training the emotion recognition model of the server by using the training material and the calibration information after the calibration information of the training material is acquired so as to update the emotion recognition model of the server;
and the sending module is used for sending the emotion recognition model of the server to the plurality of terminals as the current emotion recognition model.
According to a fifth aspect of the present invention, there is provided an electronic device, comprising a memory and a processor,
the memory is used for storing codes;
the processor is configured to execute the codes in the memory to implement a terminal-side update processing method of the emotion recognition model according to the first aspect and its alternatives, or a server-side update processing method of the emotion recognition model according to the third aspect and its alternatives.
According to a sixth aspect of the present invention, there is provided a storage medium having a program stored thereon, characterized in that the program, when executed by a processor, implements a terminal-side update processing method of the emotion recognition model relating to the first aspect and alternatives thereof, or a server-side update processing method of the emotion recognition model relating to the third aspect and alternatives thereof.
According to the emotion recognition model updating processing method and device, the electronic equipment and the medium, the emotion recognition model of the terminal can be uploaded to the server in time when the target emotion information cannot be recognized, then the server can perform secondary learning on the current image which cannot recognize the target emotion information, so that the learned emotion recognition model can further recognize the current image or similar images thereof on the basis of the original emotion recognition capability, and therefore a feasible scheme can be provided for correction and optimization of the emotion recognition model, and the recognition accuracy of the emotion recognition model can be improved.
Drawings
In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings used in the description of the embodiments or the prior art will be briefly described below, and it is obvious that the drawings in the following description are only some embodiments of the present invention, and for those skilled in the art, other drawings can be obtained according to these drawings without creative efforts.
Fig. 1 is a schematic flow chart of a terminal-side update processing method of an emotion recognition model in an embodiment of the present invention;
FIG. 2 is a flowchart illustrating step S14 according to an embodiment of the present invention;
FIG. 3 is a flow chart of a server-side update processing method of emotion recognition models in an embodiment of the present invention;
FIG. 4 is a schematic diagram of program modules of a terminal-side update processing apparatus of an emotion recognition model in an embodiment of the present invention;
FIG. 5 is a schematic diagram of program modules of a server-side update processing apparatus for emotion recognition models in an embodiment of the present invention;
fig. 6 is a schematic structural diagram of an electronic device in an embodiment of the invention.
Detailed Description
The technical solutions in the embodiments of the present invention will be clearly and completely described below with reference to the drawings in the embodiments of the present invention, and it is obvious that the described embodiments are only a part of the embodiments of the present invention, and not all of the embodiments. All other embodiments, which can be derived by a person skilled in the art from the embodiments given herein without making any creative effort, shall fall within the protection scope of the present invention.
The terms "first," "second," "third," "fourth," and the like in the description and in the claims, as well as in the drawings, if any, are used for distinguishing between similar elements and not necessarily for describing a particular sequential or chronological order. It is to be understood that the data so used is interchangeable under appropriate circumstances such that the embodiments of the invention described herein are capable of operation in sequences other than those illustrated or described herein. Furthermore, the terms "comprises," "comprising," and "having," and any variations thereof, are intended to cover a non-exclusive inclusion, such that a process, method, system, article, or apparatus that comprises a list of steps or elements is not necessarily limited to those steps or elements expressly listed, but may include other steps or elements not expressly listed or inherent to such process, method, article, or apparatus.
The technical solution of the present invention will be described in detail below with specific examples. The following several specific embodiments may be combined with each other, and details of the same or similar concepts or processes may not be repeated in some embodiments.
Fig. 1 is a schematic flow chart of a terminal-side updating processing method of an emotion recognition model in an embodiment of the present invention.
The terminal-side update processing method according to the present embodiment can be applied to any terminal or combination of terminals that use an emotion recognition model, for example, an in-vehicle terminal, a mobile phone, a tablet computer, a computer, or the like.
And if the terminal is a vehicle-mounted terminal, the current image is the image of the driver and/or the passenger acquired by the in-vehicle image acquisition component correspondingly. Further, it can be applied to functional vehicles such as shared vehicles, test driving vehicles, school buses, trucks, and the like, so that further processing can be performed based on the emotion recognition result, for example, determination of the use experience of the shared vehicles, the test driving experience of the test driving vehicles, and the like.
The terminal to which the terminal-side update processing method is applied may also be specifically a terminal determined to participate in the model learning, for example: the method can be used as a node for federal learning to participate in the training and learning of the model.
The federal learning can also be described as alliance learning, joint learning, federal machine learning and the like, and can be specifically understood as follows: a fed machine Learning or a fed Learning.
In one example, the working principle of federal learning can be as follows: the terminal as the node can download the current models from the server end respectively; wherein, part or all terminals can use respective data to train the model; then, each terminal transmits the trained model and the trained parameters or the parameters and materials for participating in training to the server; the server may update its model after it receives it.
Referring to fig. 1, a terminal-side update processing method of an emotion recognition model includes:
s11: whether the emotion recognition model of the server is updated;
if the determination result in the step S11 is yes, the step S12 may be implemented: and receiving the current emotion recognition model sent by the server.
In one embodiment, the server first sends a notification to the terminal that the characterization model is updated, so that the terminal learns that the emotion recognition model of the server is updated, which is the implementation process of step S11, and further, the terminal actively triggers the reception of the model, which is the implementation process of step S12.
In another embodiment, the server may directly send the model to the terminal after updating, and when the terminal receives the model, that is, step S11 and step S12 are performed simultaneously, it may also be understood that: step S11 may not be a step of the active determination performed by the terminal, but merely indicates that the timing or reason for receiving the model is that the model of the server is updated.
After step S12, the method may further include:
s13: whether a current image is acquired;
if the determination result in the step S13 is yes, the step S14 may be implemented: and performing emotion recognition on the current image by using the current emotion recognition model so as to determine target emotion information corresponding to the current image in a plurality of candidate emotion information.
The emotion information may be any user-defined information for representing emotion, and in one example, the emotion information may be information representing the degree of a certain emotion, such as "happy", "unhappy", and the like, or may be quantifiable information, such as a numerical value representing the happy degree, and in another example, the emotion information may also be information representing the uncharacterized degree, such as "happy", "sad", "fear", and the like. Therefore, different emotion information can be generated correspondingly in different calibration modes during training and learning.
Fig. 2 is a flowchart illustrating step S14 according to an embodiment of the present invention.
Referring to fig. 2, step S14 may include:
s141: and comparing the current image with the standard image corresponding to each candidate emotion information through the current emotion recognition model, or comparing the characteristic information in the current image with the characteristic information of each standard image, and determining the matching evaluation data of the current image and each standard image.
S142: and acquiring the target emotion information determined by the current emotion recognition model according to the matching evaluation data.
The feature information may be any information used for describing features of a human face, and different feature information may be obtained according to different ways of feature extraction and use, and in a specific example, the feature information may be, for example, information of feature points extracted from an image, information of feature lines, and the like. Correspondingly, the matching evaluation data can be understood as being used for representing the matching degree of the characteristic information in the current image and the characteristic information of the corresponding standard image by using numerical values; correspondingly, the ability to obtain the matching evaluation data can be understood as the ability of the emotion recognition model after training. Meanwhile, the matching degree can also be characterized by the antisense deviation degree, and further, the higher the deviation degree is, that is, the lower the matching degree is, the lower the deviation degree is, and the higher the matching degree is, so that the matching evaluation data can also be understood as the deviation evaluation data.
The value may be, for example, a percentage value, and in a specific example, the emotion information may be, for example: the basic emotions of happiness, depression, surprise, nausea and the like are judged by comparing the acquired current image with the standard image or the acquired human face characteristic points with the standard human face characteristic points, and since the deviation is definitely generated in the comparison, the deviation can be expressed by percentage, which is the matching evaluation data (or called deviation evaluation data).
Assume that the distribution of data at a certain time is: happy: -70%, depression: -20%, surprisingly: -7%, nausea: 13%, and the original basic emotion is compared. If the value is large, the emotion is judged. For example, with respect to the above distribution, the emotion may be identified as surprised, i.e., the target emotion information may be determined as surprised. It can be seen that the sum of the data of the distributions can reach 100%. Furthermore, the matching evaluation data, or the deviation evaluation data, may be a data distribution condition in a certain total amount of data. The data total amount is unchanged, so that the uniformity of the evaluation data scale can be guaranteed.
After step S14, the method may further include:
s15: whether the current emotion recognition model determines unique target emotion information for the current image.
If the model is designed to determine only one target emotion information, when the one target emotion information cannot be determined, it may be determined that the emotion recognition model cannot recognize the emotion, which may be understood as the result of the determination in step S15 being no; in some examples, the model may be designed to output a plurality of target emotion information, in which case, the implementation of step S15 may be understood as determining whether the unique target emotion information is determined, and if the determined target emotion information is not unique, it may be understood as the result of the determination of step S15 is no. On the contrary, if the determination result in step S15 is yes, the target emotion information can be uniquely determined without optimizing or modifying the model.
In one embodiment, the process of making the step S15 yes may include: and determining that the matching evaluation data of the two standard images with the highest matching degree are the same or the difference value is smaller than a threshold value, so that the current emotion recognition model cannot determine unique target emotion information for the current image.
For example, if the distribution of the above-mentioned values is found to be relatively similar, such as happy-40%, surprised-40%, such as happy-42%, and surprised-45%, and further the difference between the two values is small, such as 3%, and is less than a threshold of a certain example of 5%, the terminal is required to apply a similar picture and/or characteristic information to the server side to calibrate what the emotional tendency at this time is more, so as to implement the modified optimization of the model.
If the determination result in the step S15 is negative, the step S16 may be implemented: uploading the current image or the characteristic information thereof to the server so that the server can determine a training material according to the current image or the characteristic information thereof, and training an emotion recognition model of the server by using the training material so as to update the emotion recognition model of the server.
In summary, in the terminal-side update processing method of the emotion recognition model provided in this embodiment, when the emotion recognition model of the terminal cannot recognize the target emotion information, the target emotion information can be uploaded to the server in time, and then the server can perform learning again on the current image in which the target emotion information cannot be recognized, so that the learned emotion recognition model can further recognize the current image or a similar image thereof on the basis of the original emotion recognition capability.
Fig. 3 is a flowchart illustrating a server-side update processing method of an emotion recognition model in an embodiment of the present invention.
Referring to fig. 3, a server-side update processing method of an emotion recognition model, applied to a server, includes:
s21: and receiving the current image or the characteristic information thereof sent by any one target terminal in the plurality of terminals.
The current image is an image of which the current emotion recognition model cannot determine unique target emotion information when the terminal performs emotion recognition on the current image by using the current emotion recognition model; the current emotion recognition model used by the terminal is sent by the server;
s22: determining a training material according to the current image and/or the characteristic information thereof; the training material is at least one of the following: the current image, the feature information of the current image, other similar images similar to the current image and the feature information of the other similar images;
s23: whether calibration information of the training materials is obtained or not;
if the determination result in the step S23 is yes, the step S24 may be implemented: training an emotion recognition model of the server by using the training materials and the calibration information so as to update the emotion recognition model of the server;
s25: and sending the emotion recognition model of the server to the plurality of terminals as a current emotion recognition model.
The technical terms, technical features and technical effects mentioned above can be understood with reference to the related description of the terminal-side update processing method, and the repeated portions will not be described again.
The other similar images may be understood as images with a higher similarity to the current image, or images with a higher similarity between the feature information and the feature information of the current image, and may specifically be images retrieved from an external database or an internal database.
The calibration information may be artificially generated or automatically generated, and may be interpreted as emotional information for calibrating at least one of the corresponding current image, the feature information of the current image, other similar images, and the feature information of the other similar images.
In addition, since other similar images are determined according to the current image, a process of determining other similar images and/or feature information thereof according to the current image and using the other similar images and/or feature information thereof as training materials can also be understood as an embodiment of determining the training materials according to the current image.
In summary, in the server-side update processing method of the emotion recognition model provided in this embodiment, when the emotion recognition model of the terminal cannot recognize the target emotion information, the target emotion information can be uploaded to the server in time, and then the server can perform learning again on the current image in which the target emotion information cannot be recognized, so that the learned emotion recognition model can further recognize the current image or similar images thereof on the basis of the original emotion recognition capability.
Fig. 4 is a schematic diagram of program modules of a terminal-side update processing apparatus of an emotion recognition model in an embodiment of the present invention.
Referring to fig. 4, the terminal-side update processing apparatus 300 of the emotion recognition model includes:
the terminal receiving module 301 is configured to receive a current emotion recognition model sent by a server each time the emotion recognition model of the server is updated;
the identification module 302 is configured to, when a current image is acquired, perform emotion identification on the current image by using the current emotion identification model, so as to determine target emotion information corresponding to the current image from among a plurality of candidate emotion information;
an uploading module 303, configured to upload the current image or feature information thereof to the server if the current emotion recognition model cannot determine unique target emotion information for the current image, so that the server can determine the current image and/or other images similar to the current image as training materials, train an emotion recognition model in the server, and update the emotion recognition model.
Optionally, the identifying module 302 is specifically configured to:
comparing the current image with a standard image corresponding to each candidate emotion information through the current emotion recognition model, or comparing the feature information in the current image with the feature information of each standard image, and determining the matching evaluation data of the current image and each standard image; the matching evaluation data is used for representing the matching degree of the characteristic information in the current image and the characteristic information of the corresponding standard image by using a numerical value;
and acquiring the target emotion information determined by the current emotion recognition model according to the matching evaluation data.
Optionally, the identifying module 302 is further configured to:
and determining that the matching evaluation data of the two standard images with the highest matching degree are the same or the difference value is smaller than a threshold value, so that the current emotion recognition model cannot determine unique target emotion information for the current image.
Optionally, the terminal is a vehicle-mounted terminal, and the current image is an image of a driver and/or a passenger acquired by an in-vehicle image acquisition component.
In summary, in the terminal-side update processing device of the emotion recognition model provided in this embodiment, when the emotion recognition model of the terminal cannot recognize the target emotion information, the target emotion information can be uploaded to the server in time, and then the server can perform learning again for the current image in which the target emotion information cannot be recognized, so that the learned emotion recognition model can further recognize the current image or a similar image thereof on the basis of the original emotion recognition capability.
Fig. 5 is a schematic diagram of program modules of a server-side update processing apparatus of an emotion recognition model according to an embodiment of the present invention.
Referring to fig. 5, a server-side update processing apparatus 400 of an emotion recognition model includes:
a server receiving module 401, configured to receive a current image or feature information thereof sent by any one target terminal in multiple terminals, where the current image is an image in which a current emotion recognition model cannot determine unique target emotion information when the terminal performs emotion recognition on the current image by using the current emotion recognition model; the current emotion recognition model used by the terminal is sent by the server;
a material determining module 402, configured to determine a training material according to the current image and/or feature information thereof; the training material is at least one of the following: the current image, the feature information of the current image, and the feature information of other similar images similar to the current image and the other similar images
A training module 403, configured to train an emotion recognition model of the server by using the training material and the calibration information after obtaining the calibration information of the training material, so as to update the emotion recognition model of the server;
a sending module 404, configured to send the emotion recognition model of the server to the multiple terminals as a current emotion recognition model.
In summary, in the server-side update processing device of the emotion recognition model provided in this embodiment, when the emotion recognition model of the terminal cannot recognize the target emotion information, the target emotion information can be uploaded to the server in time, and then the server can perform learning again for the current image in which the target emotion information cannot be recognized, so that the learned emotion recognition model can further recognize the current image or a similar image thereof on the basis of the original emotion recognition capability.
Fig. 6 is a schematic structural diagram of an electronic device in an embodiment of the invention.
Referring to fig. 6, an electronic device 50 is provided, including:
a processor 51; and the number of the first and second groups,
a memory 52 for storing executable instructions of the processor;
wherein the processor 51 is configured to perform the above-mentioned method via execution of the executable instructions.
The processor 51 is capable of communicating with the memory 52 via a bus 53.
The present embodiments also provide a computer-readable storage medium having stored thereon a computer program which, when executed by a processor, implements the above-mentioned method.
Those of ordinary skill in the art will understand that: all or a portion of the steps of implementing the above-described method embodiments may be performed by hardware associated with program instructions. The program may be stored in a computer-readable storage medium. When executed, the program performs steps comprising the method embodiments described above; and the aforementioned storage medium includes: various media that can store program codes, such as ROM, RAM, magnetic or optical disks.
Finally, it should be noted that: the above embodiments are only used to illustrate the technical solution of the present invention, and not to limit the same; while the invention has been described in detail and with reference to the foregoing embodiments, it will be understood by those skilled in the art that: the technical solutions described in the foregoing embodiments may still be modified, or some or all of the technical features may be equivalently replaced; and the modifications or the substitutions do not make the essence of the corresponding technical solutions depart from the scope of the technical solutions of the embodiments of the present invention.
Claims (10)
1. A terminal side updating processing method of an emotion recognition model is applied to a terminal, and is characterized by comprising the following steps:
receiving a current emotion recognition model sent by a server after the emotion recognition model of the server is updated each time;
when a current image is acquired, performing emotion recognition on the current image by using the current emotion recognition model so as to determine target emotion information corresponding to the current image in a plurality of candidate emotion information;
if the current emotion recognition model cannot determine the only target emotion information for the current image, uploading the current image or the characteristic information of the current image to the server, so that the server can determine a training material according to the current image or the characteristic information of the current image, and training the emotion recognition model of the server by using the training material to update the emotion recognition model of the server.
2. The method of claim 1, wherein performing emotion recognition on the current image by using the current emotion recognition model to determine target emotion information corresponding to the current image among a plurality of candidate emotion information, comprises:
comparing the current image with a standard image corresponding to each candidate emotion information through the current emotion recognition model, or comparing the feature information in the current image with the feature information of each standard image, and determining the matching evaluation data of the current image and each standard image; the matching evaluation data is used for representing the matching degree of the characteristic information in the current image and the characteristic information of the corresponding standard image by using a numerical value;
and acquiring the target emotion information determined by the current emotion recognition model according to the matching evaluation data.
3. The method of claim 2, wherein before uploading the current image or feature information thereof to the server, further comprising:
and determining that the matching evaluation data of the two standard images with the highest matching degree are the same or the difference value is smaller than a threshold value, so that the current emotion recognition model cannot determine unique target emotion information for the current image.
4. The method according to any one of claims 1 to 3, wherein the terminal is an in-vehicle terminal, and the current image is an image of a driver and/or passenger captured by an in-vehicle image capturing means.
5. A terminal side update processing device of an emotion recognition model, comprising:
the terminal receiving module is used for receiving the current emotion recognition model sent by the server each time the emotion recognition model of the server is updated;
the identification module is used for carrying out emotion identification on the current image by using the current emotion identification model when the current image is obtained so as to determine target emotion information corresponding to the current image in a plurality of candidate emotion information;
and the uploading module is used for uploading the current image or the characteristic information thereof to the server if the current emotion recognition model cannot determine the only target emotion information aiming at the current image, so that the server can determine the current image and/or other images similar to the current image as training materials, train the emotion recognition model in the server and update the emotion recognition model.
6. A server-side updating processing method of an emotion recognition model is applied to a server and is characterized by comprising the following steps:
receiving a current image or characteristic information thereof sent by any one target terminal in a plurality of terminals, wherein the current image is an image of which the current emotion recognition model cannot determine unique target emotion information when the terminal performs emotion recognition on the current image by using the current emotion recognition model; the current emotion recognition model used by the terminal is sent by the server;
determining a training material according to the current image and/or the characteristic information thereof; the training material is at least one of the following: the current image, the feature information of the current image, other similar images similar to the current image and the feature information of the other similar images;
after the calibration information of the training material is obtained, training an emotion recognition model of the server by using the training material and the calibration information so as to update the emotion recognition model of the server;
and sending the emotion recognition model of the server to the plurality of terminals as a current emotion recognition model.
7. The method according to claim 6, wherein the terminal is an in-vehicle terminal, and the current image is an image of a driver and/or a passenger collected by an in-vehicle image collecting means.
8. A server-side update processing apparatus of an emotion recognition model, comprising:
the server receiving module is used for receiving a current image or characteristic information thereof sent by any one target terminal in a plurality of terminals, wherein the current image is an image of which the current emotion recognition model cannot determine unique target emotion information when the terminal performs emotion recognition on the current image by using the current emotion recognition model; the current emotion recognition model used by the terminal is sent by the server;
the material determining module is used for determining training materials according to the current image and/or the characteristic information thereof; the training material is at least one of the following: the current image, the feature information of the current image, and the feature information of other similar images similar to the current image and the other similar images
The training module is used for training the emotion recognition model of the server by using the training material and the calibration information after the calibration information of the training material is acquired so as to update the emotion recognition model of the server;
and the sending module is used for sending the emotion recognition model of the server to the plurality of terminals as the current emotion recognition model.
9. An electronic device, comprising a memory and a processor,
the memory is used for storing codes;
the processor is configured to execute the codes in the memory to implement the terminal-side update processing method of the emotion recognition model according to any one of claims 1 to 4 or the server-side update processing method of the emotion recognition model according to any one of claims 6 to 7.
10. A storage medium on which a program is stored, characterized in that the program, when executed by a processor, implements a terminal-side update processing method of the emotion recognition model recited in any one of claims 1 to 4, or a server-side update processing method of the emotion recognition model recited in any one of claims 6 to 7.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201911202062.XA CN110991322A (en) | 2019-11-29 | 2019-11-29 | Emotion recognition model updating method and device, electronic equipment and medium |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201911202062.XA CN110991322A (en) | 2019-11-29 | 2019-11-29 | Emotion recognition model updating method and device, electronic equipment and medium |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN110991322A true CN110991322A (en) | 2020-04-10 |
Family
ID=70088466
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201911202062.XA Pending CN110991322A (en) | 2019-11-29 | 2019-11-29 | Emotion recognition model updating method and device, electronic equipment and medium |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN110991322A (en) |
Cited By (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN112686129A (en) * | 2020-12-28 | 2021-04-20 | 湖北亿咖通科技有限公司 | Face recognition system and method |
Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20100088088A1 (en) * | 2007-01-31 | 2010-04-08 | Gianmario Bollano | Customizable method and system for emotional recognition |
| CN105894039A (en) * | 2016-04-25 | 2016-08-24 | 京东方科技集团股份有限公司 | Emotion recognition modeling method, emotion recognition method and apparatus, and intelligent device |
| CN109815817A (en) * | 2018-12-24 | 2019-05-28 | 北京新能源汽车股份有限公司 | A kind of the Emotion identification method and music method for pushing of driver |
| CN109871807A (en) * | 2019-02-21 | 2019-06-11 | 百度在线网络技术（北京）有限公司 | Face image processing process and device |
| CN110505585A (en) * | 2019-08-28 | 2019-11-26 | 上海能塔智能科技有限公司 | Data processing method, server-side, electronic equipment and the readable storage medium storing program for executing of vehicle |
-
2019
- 2019-11-29 CN CN201911202062.XA patent/CN110991322A/en active Pending
Patent Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20100088088A1 (en) * | 2007-01-31 | 2010-04-08 | Gianmario Bollano | Customizable method and system for emotional recognition |
| CN105894039A (en) * | 2016-04-25 | 2016-08-24 | 京东方科技集团股份有限公司 | Emotion recognition modeling method, emotion recognition method and apparatus, and intelligent device |
| CN109815817A (en) * | 2018-12-24 | 2019-05-28 | 北京新能源汽车股份有限公司 | A kind of the Emotion identification method and music method for pushing of driver |
| CN109871807A (en) * | 2019-02-21 | 2019-06-11 | 百度在线网络技术（北京）有限公司 | Face image processing process and device |
| CN110505585A (en) * | 2019-08-28 | 2019-11-26 | 上海能塔智能科技有限公司 | Data processing method, server-side, electronic equipment and the readable storage medium storing program for executing of vehicle |
Cited By (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN112686129A (en) * | 2020-12-28 | 2021-04-20 | 湖北亿咖通科技有限公司 | Face recognition system and method |
| CN112686129B (en) * | 2020-12-28 | 2022-05-13 | 亿咖通(湖北)技术有限公司 | Face recognition system and method |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN110288085B (en) | Data processing method, device and system and storage medium | |
| DE102021115299A1 (en) | METHOD AND DEVICE FOR CONTINUOUS FEW-SHOT LEARNING WITHOUT FORGETTING | |
| US11120308B2 (en) | Vehicle damage detection method based on image analysis, electronic device and storage medium | |
| CN112381092A (en) | Tracking method, device and computer readable storage medium | |
| CN110991322A (en) | Emotion recognition model updating method and device, electronic equipment and medium | |
| CN109190757B (en) | Task processing method, device, equipment and computer readable storage medium | |
| CN111126493B (en) | Training method and device for deep learning model, electronic equipment and storage medium | |
| CN110766086B (en) | Method and device for fusing multiple classification models based on reinforcement learning model | |
| CN108717533B (en) | Intelligent license plate recognition method and device and related equipment | |
| CN111353526A (en) | Image matching method and device and related equipment | |
| CN113409317B (en) | Defect detection method, electronic device, device and computer readable storage medium | |
| CN114148107B (en) | Component code spraying method and device and electronic equipment | |
| CN114120206A (en) | Vehicle end target detection method and device based on mobile end | |
| CN113219338B (en) | Battery degradation determination system, method, and non-transitory storage medium storing program | |
| CN114140822A (en) | Pedestrian re-identification method and device | |
| CN110807403B (en) | User identity identification method and device and electronic equipment | |
| CN114759904A (en) | Data processing method, device, equipment, readable storage medium and program product | |
| CN113065109A (en) | Man-machine recognition method and device | |
| CN110705627A (en) | Target detection method, target detection system, target detection device and readable storage medium | |
| CN113093012A (en) | Battery energy state detection method, battery energy state detection equipment, storage medium and device | |
| CN114419033A (en) | Defect identification method, defect identification device, electronic device and readable storage medium | |
| CN111009696B (en) | Storage battery maintenance method and device | |
| CN113570566B (en) | Product appearance defect development cognition detection method and related device | |
| CN111814904B (en) | Method and system for distinguishing test driving mode of automatic driving road | |
| CN116090578A (en) | Model training method, image recognition method, device, equipment and storage medium |
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 |