CN110555488A - Image sequence auditing method and system, electronic equipment and storage medium - Google Patents
Image sequence auditing method and system, electronic equipment and storage medium Download PDFInfo
- Publication number
- CN110555488A CN110555488A CN201810563849.8A CN201810563849A CN110555488A CN 110555488 A CN110555488 A CN 110555488A CN 201810563849 A CN201810563849 A CN 201810563849A CN 110555488 A CN110555488 A CN 110555488A
- Authority
- CN
- China
- Prior art keywords
- image sequence
- model
- image
- neural network
- sequence
- 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
- 238000000034 method Methods 0.000 title claims abstract description 36
- 238000013528 artificial neural network Methods 0.000 claims abstract description 11
- 238000012549 training Methods 0.000 claims description 79
- 238000003062 neural network model Methods 0.000 claims description 23
- 238000000605 extraction Methods 0.000 claims description 13
- 238000004590 computer program Methods 0.000 claims description 9
- 238000013527 convolutional neural network Methods 0.000 claims description 8
- 238000012550 audit Methods 0.000 claims description 3
- 238000012552 review Methods 0.000 abstract description 10
- 238000013135 deep learning Methods 0.000 abstract description 3
- 238000010586 diagram Methods 0.000 description 4
- 230000006870 function Effects 0.000 description 4
- 238000011156 evaluation Methods 0.000 description 3
- 239000000284 extract Substances 0.000 description 2
- 238000012986 modification Methods 0.000 description 2
- 230000004048 modification Effects 0.000 description 2
- 238000012545 processing Methods 0.000 description 2
- 230000001960 triggered effect Effects 0.000 description 2
- 238000012795 verification Methods 0.000 description 2
- 238000003491 array Methods 0.000 description 1
- 230000019771 cognition Effects 0.000 description 1
- 238000004891 communication Methods 0.000 description 1
- 230000007547 defect Effects 0.000 description 1
- 238000009795 derivation Methods 0.000 description 1
- 230000000694 effects Effects 0.000 description 1
- 238000005516 engineering process Methods 0.000 description 1
- 238000001914 filtration Methods 0.000 description 1
- 238000007689 inspection Methods 0.000 description 1
- 230000003287 optical effect Effects 0.000 description 1
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/04—Architecture, e.g. interconnection topology
- G06N3/045—Combinations of networks
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V30/00—Character recognition; Recognising digital ink; Document-oriented image-based pattern recognition
- G06V30/10—Character recognition
- G06V30/19—Recognition using electronic means
- G06V30/192—Recognition using electronic means using simultaneous comparisons or correlations of the image signals with a plurality of references
- G06V30/194—References adjustable by an adaptive method, e.g. learning
Abstract
The invention discloses an image sequence auditing method and system, electronic equipment and a storage medium. The auditing method comprises the following steps: establishing an image sequence model based on a neural network; acquiring an image sequence to be audited; and extracting the image characteristics of each image in the image sequence, inputting the image characteristics into the image sequence model, and calculating the confidence coefficient that the image sequence meets the release requirement. According to the invention, manual review is replaced by the image sequence model based on deep learning, so that automatic review of the image sequence is realized, the accuracy is greatly improved, and the labor cost is reduced.
Description
Technical Field
The present invention relates to the field of computer technologies, and in particular, to a method and a system for auditing an image sequence, an electronic device, and a storage medium.
background
In order to meet the relevant regulations, before the images are released on the internet site, the images need to be checked. For example, on an e-commerce or third-party evaluation website, the correlation between pictures and characters, the quality of the pictures, whether illegal elements exist, and the correlation sequence of multiple related pictures need to be checked. Different web sites have different requirements.
At present, there are two main ways of image review:
(1) and (4) pure manual auditing, namely, auditing the single pictures and the picture sequence inspection one by special service personnel according to auditing rules. The mode has high labor cost, low efficiency and high error rate;
(2) and semi-manual review, namely, simply classifying and filtering all the images by a classifier or a detector, and manually reviewing whether the image sequence is reasonable or not. Although this method can screen out single picture which is not satisfactory, the auditing of the image sequence still needs to be completed manually, and the workload is still large.
Disclosure of Invention
the invention provides an image sequence auditing method and system, electronic equipment and a storage medium, aiming at overcoming the defect that the image sequence auditing cannot be realized in the prior art.
The invention solves the technical problems through the following technical scheme:
An auditing method of a sequence of images, the auditing method comprising:
Establishing an image sequence model based on a neural network;
Acquiring an image sequence to be audited;
And extracting the image characteristics of each image in the image sequence, inputting the image characteristics into the image sequence model, and calculating the confidence coefficient that the image sequence meets the release requirement.
preferably, the auditing method further comprises:
Judging whether the confidence of the image sequence is greater than a threshold value;
if the judgment is negative, prompting the complaint information;
when a complaint instruction is received, rechecking a target image sequence with the reliability less than or equal to a threshold value;
When a rechecking failure instruction is received, prompting that information is not issued;
And when a rechecking passing instruction is received, issuing the target image sequence.
Preferably, the establishing of the image sequence model based on the neural network specifically includes:
acquiring a marked image sequence as a first training sample;
Extracting image characteristics of each image in the first training sample and sequentially inputting the image characteristics into a neural network model;
The output parameter of the neural network model is confidence;
and training the neural network model in an end-to-end training mode, and determining parameters of the neural network model to obtain the image sequence model.
Preferably, after the step of reviewing the target image sequence, the method further includes:
Marking the target image sequence and using the target image sequence as a second training sample;
And training the image sequence model according to the second training sample so as to update the parameters of the model.
Preferably, the extracting the image features of each image specifically includes:
And extracting the image features based on a deep convolutional neural network model.
The invention further provides an electronic device, which includes a memory, a processor, and a computer program stored on the memory and executable on the processor, and when the processor executes the computer program, the processor implements the method for auditing the image sequence according to any one of the above items.
The invention also provides a computer-readable storage medium, on which a computer program is stored, which computer program, when being executed by a processor, carries out the steps of the method for auditing a sequence of images according to any one of the preceding claims.
The invention also provides an auditing system for image sequences, which comprises:
The model establishing module is used for establishing an image sequence model based on a neural network;
The image sequence acquisition model is used for acquiring an image sequence to be audited;
The characteristic extraction module is used for extracting the image characteristics of each image in the image sequence and inputting the image characteristics into the image sequence model;
The image sequence model is used for calculating the confidence degree that the image sequence meets the release requirement.
Preferably, the auditing system further includes: the system comprises a judgment module, a prompt module, a rechecking module and a release module;
the judgment model is used for judging whether the confidence coefficient of the image sequence is greater than a threshold value;
when the judgment of the judgment model is negative, calling the prompt module to prompt the complaint information;
The rechecking module is used for rechecking the target image sequence with the credibility less than or equal to the threshold value when receiving the complaint instruction, calling the prompting module to prompt that information is not issued when receiving a rechecking non-passing instruction, and calling the issuing module to issue the target image sequence when receiving a rechecking passing instruction.
Preferably, the model building module specifically includes:
A sample acquisition unit for acquiring the marked image sequence as a first training sample;
The characteristic extraction unit is used for extracting the image characteristics of each image in the first training sample and sequentially inputting the image characteristics into a neural network model;
The output parameter of the neural network model is confidence;
and the model training unit is used for training the neural network model in an end-to-end training mode and determining parameters of the neural network model to obtain the image sequence model.
Preferably, the sample acquiring unit is further configured to mark the target image sequence as a second training sample;
The model training unit is further configured to train the image sequence model according to the second training sample to update parameters of the model.
Preferably, the feature extraction module is specifically configured to extract the image features based on a deep convolutional neural network model.
the positive progress effects of the invention are as follows: according to the invention, manual review is replaced by the image sequence model based on deep learning, so that automatic review of the image sequence is realized, the accuracy is greatly improved, and the labor cost is reduced.
Drawings
fig. 1 is a flowchart of an auditing method for an image sequence according to embodiment 1 of the present invention.
Fig. 2 is a flowchart illustrating specific steps of building the image sequence model in fig. 1.
fig. 3 is a flowchart of an auditing method for an image sequence according to embodiment 2 of the present invention.
Fig. 4 is a schematic diagram of a hardware structure of an electronic device according to embodiment 3 of the present invention.
Fig. 5 is a schematic block diagram of an audit system of an image sequence according to embodiment 5 of the present invention.
Detailed Description
The invention is further illustrated by the following examples, which are not intended to limit the scope of the invention.
example 1
The embodiment provides an auditing method of an image sequence based on deep learning, which can automatically and accurately identify whether the image sequence meets the release requirements of an internet website. A sequence of images is also a set of images with ordering rules. As shown in fig. 1, the auditing method includes the following steps:
Step 101, establishing an image sequence model based on a neural network.
as shown in fig. 2, step 101 specifically includes:
Step 101-1, a marked image sequence is obtained as a first training sample.
step 101-1 is also to construct an initial sample set of the trained neural network. The marked image sequence is obtained from an internet site as an initial sample. For a new service without historical data, some image sequences can be manually marked to serve as training samples, namely some image sequences are collected and marked by adopting an auditing standard established by the service.
The labels of the marks are, for example, "go up", "down". The online, that is, the image sequence conforms to the internet release standard, and the compliance with the release standard here includes two layers of meanings: on one hand, each image is clearer and has no elements forbidden by yellow-related, advertising and other services; on the other hand, it is reasonable to see the images in sequence in each group of image sequences, i.e. the displayed images are in the same order as the conventional cognition and have a common theme, but the contents are different. I.e. the training sample should contain information on whether the individual images are compliant or not, in addition to the label information. The lower line, i.e. the image sequence, does not comply with the internet release standard.
Because the data volume of the sample set is usually large, in order to ensure the speed of data access for training and avoid repeated downloading during model training, the image data should be stored on a local or network disk accessible to the machine where the training is located after the sample set is constructed.
And 101-2, extracting the image characteristics of each image in the first training sample and sequentially inputting the image characteristics into a neural network model.
wherein, the output parameter of the neural network model is confidence.
specifically, in step 101-2, image features are extracted based on a trained deep convolutional neural network model.
101-3, training the neural network model by adopting an end-to-end training mode, and determining parameters of the neural network model.
Step 101-4, determining an image sequence model.
wherein, the evaluation function of the model training adopts cross entropy. In the embodiment, because the training sample implicitly contains the information whether a single image is in compliance, only one model needs to be deployed in the auditing process, and a plurality of classifiers or detectors are not needed, so that the efficiency is greatly improved.
And 102, acquiring an image sequence to be audited.
The image sequence to be audited is also the image sequence to be uploaded to the internet site by the user.
And 103, extracting the image characteristics of each image in the image sequence, inputting the image sequence model, and calculating the confidence coefficient that the image sequence meets the release requirement.
Similarly, in step 103, a deep convolutional neural network model is also used to extract the image features of each image in the image sequence to be reviewed.
and 104, judging whether the confidence of the image sequence is greater than a threshold value.
If it is determined that the image sequence is compliant with the image distribution request of the internet site, step 105 is executed.
If not, the image sequence is not in accordance with the image publishing requirement of the internet site, and step 106 is executed.
it should be noted that the threshold may be set according to actual requirements, and in order to ensure the accuracy of online image review, the determination threshold of the sequence model should be a high threshold, which is generally set to be above 0.9.
Step 105, the image sequence is released.
And step 106, prompting complaint information.
the complaint information comprises the information that the image sequence cannot be approved and whether the complaint prompt is given, and a user can select complaint or modify or re-upload the image sequence according to the complaint information.
and step 107, whether a complaint instruction is received.
If the complaint instruction lifted by the user is not received within a preset time period (the time can be set by the user), the group of image sequences is off-line by default.
If the complaint instruction is received within the preset time period, which indicates that the user proposes a complaint, step 108 is executed.
If the user re-uploads the image sequence, step 102 is performed on the re-uploaded image sequence.
and 108, rechecking the target image sequence with the reliability less than or equal to the threshold value.
After the user proposes the complaint, the target image sequence is pushed to a service staff for manual review (manual intervention), and the service staff carefully considers according to the service rule and then gives final judgment, namely whether the image sequence meets the release requirement or not. In the embodiment, only when the confidence of the image sequence is judged to be less than or equal to the threshold value by the image sequence model and the user proposes a complaint, the intervention and the verification of the service personnel are needed, so that the labor cost is greatly reduced, and meanwhile, the working intensity of the service personnel is reduced, and the condition that the accuracy of the judgment is influenced by fatigue can be avoided.
And step 109, judging whether the recheck is passed or not.
If the rechecking result is that the rechecking is not passed, namely a rechecking not-passing instruction is received, prompting that no information is issued, namely the target image sequence is off-line; if the double check result is that the double check is passed, which indicates that the double check passed instruction is received, step 105 is executed.
Example 2
Embodiment 2 is basically the same as embodiment 1, except that the auditing method of this embodiment also performs tuning training on the image sequence model. As shown in fig. 3, after step 108, the auditing method of this embodiment further includes:
And step 110, marking the target image sequence as a second training sample.
And the target image sequence is the image sequence which is proposed by the user after the judgment of the image sequence model, the service personnel can mark the image sequence after rechecking, and the marked target image sequence is added into a difficult sample database to be used as a second training sample for tuning and training the model. The second training sample is integrated into the initial sample set and automatically downloaded. During the model training, the number of samples for each training can be set, for example, 5000 samples, when the number of samples exceeds 5000, the training process is triggered, otherwise, the number of samples is continuously detected.
And 111, training the image sequence model according to the second training sample, and updating parameters of the model.
the second training sample is often an image sequence with disputed, the model is easy to judge wrong samples, the second training sample is added to update the model parameters, the accuracy and robustness of model judgment can be improved, and thus a virtuous circle is formed.
Example 3
fig. 4 is a schematic structural diagram of an electronic device provided in embodiment 3 of the present invention, and shows a block diagram of an exemplary electronic device 30 suitable for implementing an embodiment of the present invention. The electronic device 30 shown in fig. 4 is only an example, and should not bring any limitation to the functions and the scope of use of the embodiment of the present invention.
as shown in fig. 4, the electronic device 30 may be embodied in the form of a general purpose computing device, which may be, for example, a server device. The components of the electronic device 30 may include, but are not limited to: the at least one processor 31, the at least one memory 32, and a bus 33 connecting the various system components (including the memory 32 and the processor 31).
The bus 33 includes a data bus, an address bus, and a control bus.
the memory 32 may include volatile memory, such as Random Access Memory (RAM)321 and/or cache memory 322, and may further include Read Only Memory (ROM) 323.
Memory 32 may also include a program/utility 325 having a set (at least one) of program modules 324, such program modules 324 including, but not limited to: an operating system, one or more application programs, other program modules, and program data, each of which, or some combination thereof, may comprise an implementation of a network environment.
The processor 31 executes a computer program stored in the memory 32, thereby executing various functional applications and data processing, such as an auditing method for an image sequence provided in embodiment 1 or 2 of the present invention.
The electronic device 30 may also communicate with one or more external devices 34 (e.g., keyboard, pointing device, etc.). Such communication may be through input/output (I/O) interfaces 35. Also, the model-generating electronic device 30 may communicate with one or more networks (e.g., a Local Area Network (LAN), a Wide Area Network (WAN), and/or a public network, such as the Internet) via the network adapter 36. As shown, network adapter 36 communicates with the other modules of model-generated electronic device 30 via bus 33. It should be appreciated that although not shown in the figures, other hardware and/or software modules may be used in conjunction with the model-generating electronic device 30, including but not limited to: microcode, device drivers, redundant processors, external disk drive arrays, RAID (disk array) systems, tape drives, and data backup storage systems, etc.
It should be noted that although in the above detailed description several units/modules or sub-units/modules of the electronic device are mentioned, such a division is merely exemplary and not mandatory. Indeed, the features and functionality of two or more of the units/modules described above may be embodied in one unit/module according to embodiments of the invention. Conversely, the features and functions of one unit/module described above may be further divided into embodiments by a plurality of units/modules.
Example 4
The present embodiment provides a computer-readable storage medium on which a computer program is stored, which program, when being executed by a processor, implements the steps of the method for auditing a sequence of images provided in embodiment 1 or 2.
More specific examples, among others, that the readable storage medium may employ may include, but are not limited to: a portable disk, a hard disk, random access memory, read only memory, erasable programmable read only memory, optical storage device, magnetic storage device, or any suitable combination of the foregoing.
in a possible implementation, the invention can also be implemented in the form of a program product comprising program code for causing a terminal device to perform the steps of implementing the method for auditing a sequence of images as provided in embodiment 1 or 2, when the program product is run on the terminal device.
where program code for carrying out the invention is written in any combination of one or more programming languages, the program code may be executed entirely on the user device, partly on the user device, as a stand-alone software package, partly on the user device and partly on a remote device or entirely on the remote device.
Example 5
As shown in fig. 5, the system for checking an image sequence of the present embodiment includes: the image processing system comprises a model establishing module 1, an image sequence obtaining module 2, a feature extraction module 3, a judgment module 4, a prompt module 5, a rechecking module 6 and a release module 7.
The model building module 1 is used for building an image sequence model based on a neural network. The input parameters of the image sequence model are image features, and the output parameters are confidence degrees.
The model building module 1 specifically includes: a sample acquisition unit 11, a feature extraction unit 12, and a model training unit 13.
The sample acquiring unit 11 is configured to acquire the marked image sequence as a first training sample. Namely, an initial sample set for training the neural network is constructed. The sample acquiring unit 11 acquires the marked image sequence as an initial sample from the internet site. For a new service without historical data, some image sequences can be manually marked to serve as training samples, namely some image sequences are collected and marked by adopting an auditing standard established by the service. The training sample should include information on whether the individual images are compliant in addition to the label information.
because the data volume of the sample set is usually large, in order to ensure the speed of data access for training and avoid repeated downloading during model training, the image data should be stored on a local or network disk accessible to the machine where the training is located after the sample set is constructed.
the feature extraction unit 12 is configured to extract image features of each image in the first training sample and sequentially input the image features into the neural network model. Specifically, the feature extraction unit 12 extracts image features based on a deep convolutional neural network model.
The model training unit 13 is configured to train the neural network model in an end-to-end training manner, and determine parameters of the neural network model to obtain an image sequence model. Wherein, the evaluation function of the model training adopts cross entropy. In the embodiment, because the training sample implicitly contains the information whether a single image is in compliance, only one model needs to be deployed in the auditing process, and a plurality of classifiers or detectors are not needed, so that the efficiency is greatly improved.
The image sequence acquisition model 2 is used for acquiring an image sequence to be examined. The image sequence to be audited is also the image sequence to be uploaded to the internet site by the user.
The feature extraction module 3 is used for extracting image features of each image in the image sequence, inputting the image sequence model, and calculating the confidence coefficient that the image sequence meets the release requirement. Similar to the feature extraction unit 12, the feature extraction module 3 also extracts image features based on a deep convolutional neural network model.
the judgment model 4 is used for judging whether the confidence of the image sequence is larger than a threshold value. If the image sequence meets the image publishing requirement of the internet website, the judging module 4 calls the publishing module 7 to publish the image sequence. If the judgment result is no, the image sequence is not in accordance with the image release requirement of the internet website, and the judgment model 4 calls a prompt module 5 to prompt the complaint information. The complaint information comprises that the image sequence cannot be audited and whether the complaint is prompted, and a user can select complaint or modify or re-upload the image sequence according to the complaint information. If the user selects a complaint, a complaint instruction is generated.
It should be noted that the threshold may be set according to actual requirements, and in order to ensure the accuracy of online image review, the judgment threshold of the image sequence model should be a high threshold, which is generally set to be above 0.9.
And the rechecking module 6 is used for rechecking the target image sequence with the reliability less than or equal to the threshold value when the complaint instruction is received. And the rechecking is to push the image sequence with the confidence coefficient less than or equal to the threshold value to a service person for manual rechecking (manual intervention), and the service person carefully considers according to the service rule and then gives final judgment, namely whether the image sequence meets the release requirement. If the issue requirement is met, generating a recheck passing instruction; and if the issue requirement is not met, generating a double-check non-passing instruction.
When receiving a rechecking failure instruction, the rechecking module 6 calls the prompting module 5 to prompt that information is not issued, and when receiving a rechecking failure instruction, calls the issuing module 7 to issue a target image sequence.
in the embodiment, only when the confidence of the image sequence is judged to be less than or equal to the threshold value by the image sequence model and the user proposes a complaint, the intervention and the verification of the service personnel are needed, so that the labor cost is greatly reduced, and meanwhile, the working intensity of the service personnel is reduced, and the condition that the accuracy of the judgment is influenced by fatigue can be avoided.
in order to improve the accuracy and robustness of the model judgment, the auditing system in this embodiment also performs tuning training on the image sequence model in the process of auditing the images. The specific implementation mode is as follows:
The sample acquiring unit 11 is further configured to combine the target image sequence as a second training sample. Namely, the target image sequence which is proposed by the user after the image sequence model is judged is used as a second training sample, the service personnel can mark the target image sequence after rechecking, and the marked target image sequence is used for tuning training of the model. The second training sample is integrated into the initial sample set and automatically downloaded. During the model training, the number of samples for each training can be set, for example, 5000 samples, when the number of samples exceeds 5000, the training process is triggered, otherwise, the number of samples is continuously detected.
The model training unit 13 is further configured to train the image sequence model according to the second training samples to update parameters of the model.
The second training sample is often an image sequence with disputed, the model is easy to judge wrong samples, the second training sample is added to update the model parameters, the accuracy and robustness of model judgment can be improved, and thus a virtuous circle is formed.
in this embodiment, model training and online image review are isolated, and only the derivation (inference) part needs to be reserved for online model deployment, while sufficient computing resources should be ensured so as not to affect user experience.
while specific embodiments of the invention have been described above, it will be appreciated by those skilled in the art that this is by way of example only, and that the scope of the invention is defined by the appended claims. Various changes and modifications to these embodiments may be made by those skilled in the art without departing from the spirit and scope of the invention, and these changes and modifications are within the scope of the invention.
Claims (12)
1. An auditing method for a sequence of images, the auditing method comprising:
Establishing an image sequence model based on a neural network;
Acquiring an image sequence to be audited;
and extracting the image characteristics of each image in the image sequence, inputting the image characteristics into the image sequence model, and calculating the confidence coefficient that the image sequence meets the release requirement.
2. An auditing method for a sequence of images according to claim 1, the auditing method further comprising:
Judging whether the confidence of the image sequence is greater than a threshold value;
If the judgment is negative, prompting the complaint information;
when a complaint instruction is received, rechecking a target image sequence with the reliability less than or equal to a threshold value;
When a rechecking failure instruction is received, prompting that information is not issued;
And when a rechecking passing instruction is received, issuing the target image sequence.
3. An auditing method for an image sequence according to claim 2, characterized in that the establishing of an image sequence model based on a neural network specifically comprises:
acquiring a marked image sequence as a first training sample;
Extracting image characteristics of each image in the first training sample and sequentially inputting the image characteristics into a neural network model;
The output parameter of the neural network model is confidence;
and training the neural network model in an end-to-end training mode, and determining parameters of the neural network model to obtain the image sequence model.
4. An auditing method for an image sequence according to claim 3, following the step of reviewing the target image sequence, further comprising:
marking the target image sequence and using the target image sequence as a second training sample;
And training the image sequence model according to the second training sample so as to update the parameters of the model.
5. An auditing method for an image sequence according to any one of claims 1 to 4 characterised in that extracting the image features of each image specifically comprises:
and extracting the image features based on a deep convolutional neural network model.
6. an electronic device comprising a memory, a processor and a computer program stored on the memory and executable on the processor, wherein the processor implements the method of auditing a sequence of images according to any one of claims 1 to 5 when executing the computer program.
7. a computer-readable storage medium, on which a computer program is stored which, when being executed by a processor, carries out the steps of the method of auditing of image sequences according to any one of claims 1 to 5.
8. an audit system of a sequence of images, the audit system comprising:
The model establishing module is used for establishing an image sequence model based on a neural network;
The image sequence acquisition model is used for acquiring an image sequence to be audited;
The characteristic extraction module is used for extracting the image characteristics of each image in the image sequence and inputting the image characteristics into the image sequence model;
the image sequence model is used for calculating the confidence degree that the image sequence meets the release requirement.
9. An auditing system for a sequence of images according to claim 8, the auditing system further comprising: the system comprises a judgment module, a prompt module, a rechecking module and a release module;
The judgment model is used for judging whether the confidence coefficient of the image sequence is greater than a threshold value;
when the judgment of the judgment model is negative, calling the prompt module to prompt the complaint information;
The rechecking module is used for rechecking the target image sequence with the credibility less than or equal to the threshold value when receiving the complaint instruction, calling the prompting module to prompt that information is not issued when receiving a rechecking non-passing instruction, and calling the issuing module to issue the target image sequence when receiving a rechecking passing instruction.
10. an auditing system for an image sequence according to claim 9, where the model building module specifically comprises:
a sample acquisition unit for acquiring the marked image sequence as a first training sample;
The characteristic extraction unit is used for extracting the image characteristics of each image in the first training sample and sequentially inputting the image characteristics into a neural network model;
the output parameter of the neural network model is confidence;
And the model training unit is used for training the neural network model in an end-to-end training mode and determining parameters of the neural network model to obtain the image sequence model.
11. An auditing system for an image sequence according to claim 10, where the sample acquisition unit is further arranged to label the target image sequence as a second training sample;
The model training unit is further configured to train the image sequence model according to the second training sample to update parameters of the model.
12. An auditing system for an image sequence according to any of claims 8 to 11 characterised in that the feature extraction module is particularly operable to extract the image features based on a deep convolutional neural network model.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201810563849.8A CN110555488A (en) | 2018-06-04 | 2018-06-04 | Image sequence auditing method and system, electronic equipment and storage medium |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201810563849.8A CN110555488A (en) | 2018-06-04 | 2018-06-04 | Image sequence auditing method and system, electronic equipment and storage medium |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN110555488A true CN110555488A (en) | 2019-12-10 |
Family
ID=68735942
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201810563849.8A Pending CN110555488A (en) | 2018-06-04 | 2018-06-04 | Image sequence auditing method and system, electronic equipment and storage medium |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN110555488A (en) |
Cited By (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN111080444A (en) * | 2019-12-31 | 2020-04-28 | 中国银行股份有限公司 | Information auditing method and device |
| CN112184143A (en) * | 2020-09-07 | 2021-01-05 | 支付宝(杭州)信息技术有限公司 | Model training method, device and equipment in compliance audit rule |
| CN112866739A (en) * | 2021-01-21 | 2021-05-28 | 商客通尚景科技（上海）股份有限公司 | Method for playing live photos in turn in real time |
| CN114189709A (en) * | 2021-11-12 | 2022-03-15 | 北京天眼查科技有限公司 | Method and device for auditing video, storage medium and electronic equipment |
Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN102547794A (en) * | 2012-01-12 | 2012-07-04 | 郑州金惠计算机系统工程有限公司 | Identification and supervision platform for pornographic images and videos and inappropriate contents on wireless application protocol (WAP)-based mobile media |
| JP2015149030A (en) * | 2014-02-07 | 2015-08-20 | 国立大学法人名古屋大学 | Video content violence degree evaluation device, video content violence degree evaluation method, and video content violence degree evaluation program |
| CN105844239A (en) * | 2016-03-23 | 2016-08-10 | 北京邮电大学 | Method for detecting riot and terror videos based on CNN and LSTM |
| CN106776842A (en) * | 2016-11-28 | 2017-05-31 | 腾讯科技（上海）有限公司 | Multi-medium data detection method and device |
| CN107895172A (en) * | 2017-11-03 | 2018-04-10 | 北京奇虎科技有限公司 | Utilize the method, apparatus and computing device of image information detection anomalous video file |
-
2018
- 2018-06-04 CN CN201810563849.8A patent/CN110555488A/en active Pending
Patent Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN102547794A (en) * | 2012-01-12 | 2012-07-04 | 郑州金惠计算机系统工程有限公司 | Identification and supervision platform for pornographic images and videos and inappropriate contents on wireless application protocol (WAP)-based mobile media |
| JP2015149030A (en) * | 2014-02-07 | 2015-08-20 | 国立大学法人名古屋大学 | Video content violence degree evaluation device, video content violence degree evaluation method, and video content violence degree evaluation program |
| CN105844239A (en) * | 2016-03-23 | 2016-08-10 | 北京邮电大学 | Method for detecting riot and terror videos based on CNN and LSTM |
| CN106776842A (en) * | 2016-11-28 | 2017-05-31 | 腾讯科技（上海）有限公司 | Multi-medium data detection method and device |
| CN107895172A (en) * | 2017-11-03 | 2018-04-10 | 北京奇虎科技有限公司 | Utilize the method, apparatus and computing device of image information detection anomalous video file |
Non-Patent Citations (1)
| Title |
|---|
| 唐爽: "基于深度神经网络的微表情识别", 《电子技术与软件工程》, 31 December 2017 (2017-12-31), pages 3 * |
Cited By (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN111080444A (en) * | 2019-12-31 | 2020-04-28 | 中国银行股份有限公司 | Information auditing method and device |
| CN111080444B (en) * | 2019-12-31 | 2023-08-29 | 中国银行股份有限公司 | Information auditing method and device |
| CN112184143A (en) * | 2020-09-07 | 2021-01-05 | 支付宝(杭州)信息技术有限公司 | Model training method, device and equipment in compliance audit rule |
| CN112184143B (en) * | 2020-09-07 | 2022-04-29 | 支付宝(杭州)信息技术有限公司 | Model training method, device and equipment in compliance audit rule |
| CN112866739A (en) * | 2021-01-21 | 2021-05-28 | 商客通尚景科技（上海）股份有限公司 | Method for playing live photos in turn in real time |
| CN114189709A (en) * | 2021-11-12 | 2022-03-15 | 北京天眼查科技有限公司 | Method and device for auditing video, storage medium and electronic equipment |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN110555488A (en) | Image sequence auditing method and system, electronic equipment and storage medium | |
| US20190333118A1 (en) | Cognitive product and service rating generation via passive collection of user feedback | |
| US20230199013A1 (en) | Attack situation visualization device, attack situation visualization method and recording medium | |
| CN111767228B (en) | Interface testing method, device, equipment and medium based on artificial intelligence | |
| CN110581898A (en) | internet of things data terminal system based on 5G and edge calculation | |
| US20220148113A1 (en) | Machine learning modeling for protection against online disclosure of sensitive data | |
| CN108573268A (en) | Image-recognizing method and device, image processing method and device and storage medium | |
| CN110019163A (en) | Method, system, equipment and the storage medium of prediction, the recommendation of characteristics of objects | |
| CN111178147B (en) | Screen crushing and grading method, device, equipment and computer readable storage medium | |
| CN110288755A (en) | The invoice method of inspection, server and storage medium based on text identification | |
| CN112613569A (en) | Image recognition method, and training method and device of image classification model | |
| CN109597902A (en) | Picture examination method, apparatus, equipment and storage medium | |
| CN106446123A (en) | Webpage verification code element identification method | |
| CN113094287B (en) | Page compatibility detection method, device, equipment and storage medium | |
| CN107533636A (en) | Pre-matching prediction for pattern test | |
| CN112132220A (en) | Self-training method, system, device, electronic equipment and storage medium | |
| CN111382383A (en) | Method, device, medium and computer equipment for determining sensitive type of webpage content | |
| CN109783713A (en) | A kind of dynamic website classification method, system, equipment and medium | |
| CN111126503B (en) | Training sample generation method and device | |
| CN113419951A (en) | Artificial intelligence model optimization method and device, electronic equipment and storage medium | |
| CN112749978A (en) | Detection method, apparatus, device, storage medium, and program product | |
| KR102217092B1 (en) | Method and apparatus for providing quality information of application | |
| CN112434650A (en) | Multi-spectral image building change detection method and system | |
| CN110912918A (en) | Page repairing method and device | |
| CN111225297A (en) | Broadband passive optical network port resource remediation method and system |
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 |