CN115393966A - Dispute mediation data processing method and system based on credit supervision - Google Patents
Dispute mediation data processing method and system based on credit supervision Download PDFInfo
- Publication number
- CN115393966A CN115393966A CN202211326958.0A CN202211326958A CN115393966A CN 115393966 A CN115393966 A CN 115393966A CN 202211326958 A CN202211326958 A CN 202211326958A CN 115393966 A CN115393966 A CN 115393966A
- Authority
- CN
- China
- Prior art keywords
- feature map
- neural network
- convolutional neural
- map
- credit
- 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.)
- Granted
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
- G06V40/30—Writer recognition; Reading and verifying signatures
- G06V40/33—Writer recognition; Reading and verifying signatures based only on signature image, e.g. static signature recognition
-
- 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
- G06N3/08—Learning methods
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
- G06Q50/00—Systems or methods specially adapted for specific business sectors, e.g. utilities or tourism
- G06Q50/10—Services
- G06Q50/18—Legal services; Handling legal documents
- G06Q50/182—Alternative dispute resolution
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V10/00—Arrangements for image or video recognition or understanding
- G06V10/70—Arrangements for image or video recognition or understanding using pattern recognition or machine learning
- G06V10/764—Arrangements for image or video recognition or understanding using pattern recognition or machine learning using classification, e.g. of video objects
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V10/00—Arrangements for image or video recognition or understanding
- G06V10/70—Arrangements for image or video recognition or understanding using pattern recognition or machine learning
- G06V10/82—Arrangements for image or video recognition or understanding using pattern recognition or machine learning using neural networks
Abstract
The application relates to the field of intelligent verification, and particularly discloses a dispute mediation data processing method and system based on credit supervision.
Description
Technical Field
The present application relates to the field of intelligent verification, and more particularly, to a dispute mediation data processing method and system based on credit supervision.
Background
The mediation system is an important legal system and plays an important role in the aspect of multi-dispute resolution, but because the mediation is the right and obligation of disputes of two parties or a plurality of parties, the mediation method is a voluntarily achieved dispute resolution under the supervision of national courts, mediation committees of people and related organizations, and has no mandatory enforcement force. Therefore, the function of mediating in resolving disputes is not really played, and many disputes which are distributed from the court are returned to the court, so that the disputes are not resolved as soon as possible, and the complaints of the parties are increased.
Promoting the establishment of the social credit system and the application of credit in social treatment, credit is just beginning to become an important factor in basic social treatment. The credit mediation is based on the integrity principle, and the credit is warned to practise the noco by the person concerned with the loss of credit through credit warning, combined punishment and the like in the mediation process, so that the social dispute is flexibly resolved, and the basic level social management is promoted in a non-complaint mode.
Therefore, the mediator is integrated with a social credit supervision system in dispute mediation work, adopts a credit mediation method of ' prior credit promise ', ' supervision for the prior affairs and ' later loss of credit and punishment ', utilizes an informatization technology to carry out digital management on dispute information and mediation process, and links up credible deposit certificates on evidence materials such as ' credit promise ' and ' mediation agreement ' signed by parties in the mediation process, so that traceability of the whole mediation process, untreatable evidence materials and safe compliance processing of dispute mediation information are realized.
However, before a trustable certificate is attached to a proof material such as "letter of promise" and "mediation agreement", although a block link network can guarantee that the proof material has non-tamper-resistance, it cannot guarantee authenticity of the proof before the attachment, for example, a problem of counterfeit signature occurs.
Therefore, in order to improve the effectiveness of dispute resolution based on credit supervision, it is desirable to intelligently check the evidence material to be uploaded to ensure the correctness of the uploaded data.
Disclosure of Invention
The present application is proposed to solve the above-mentioned technical problems. The embodiment of the application provides a dispute mediation data processing method and a dispute mediation data processing system based on credit supervision, which are used for intelligently checking evidence materials to be uploaded through comparison between signatures in the evidence materials and reference signatures so as to ensure the authenticity of the evidence materials. And then the accuracy of intelligent verification of the evidence materials to be uploaded is improved so as to ensure the authenticity of the uploaded data.
According to one aspect of the application, a dispute mediation data processing method based on credit supervision is provided, and comprises the following steps:
acquiring evidence materials to be uploaded to a block chain network;
intercepting a user signature region from the evidence material;
processing the user signature region using a SLIC algorithm to generate a superpixel user signature region image;
acquiring a reference signature image;
passing the super-pixel user signature area image and the reference signature image through a twin network comprising a first convolutional neural network and a second convolutional neural network to obtain a detection feature map and a reference feature map, wherein the first convolutional neural network and the second convolutional neural network have the same network structure;
calculating a differential feature map between the detection feature map and the reference feature map; and
and passing the differential feature map through a classifier to obtain a classification result, wherein the classification result is used for indicating whether the signature in the evidence material is real or not.
In the credit supervision-based dispute resolution data processing method, the passing the superpixel user signature area image and the reference signature image through a twin network including a first convolutional neural network and a second convolutional neural network to obtain a detection feature map and a reference feature map includes: depth convolution encoding the super-pixel user signature area image using a plurality of convolutional layers of the first convolutional neural network to output a depth detection feature map from a last layer of the plurality of convolutional layers; inputting the depth detection feature map into a first spatial attention module of the first convolutional neural network to obtain a first spatial attention map; and calculating the position-based multiplication of the depth detection feature map and the first spatial attention map to obtain the detection feature map.
In the credit supervision-based dispute resolution data processing method, the inputting the depth detection feature map into a first spatial attention module of the first convolutional neural network to obtain a first spatial attention map includes: convolution coding the depth detection feature map using convolution layers of the first spatial attention module to obtain a first spatial attention score map; and non-linearly activating the first spatial attention score map using a Softmax activation function to obtain the first spatial attention score map.
In the credit supervision-based dispute resolution data processing method, the passing the superpixel user signature area image and the reference signature image through a twin network including a first convolutional neural network and a second convolutional neural network to obtain a detection feature map and a reference feature map includes: depth convolution encoding the super-pixel user signature region image using multi-layer convolution layers of the second convolutional neural network to output a depth reference feature map from a last layer of the multi-layer convolution layers; inputting the depth reference feature map into a second spatial attention module of the second convolutional neural network to obtain a second spatial attention map; and calculating the position-based multiplication of the depth reference feature map and the second spatial attention map to obtain the reference feature map.
In the credit supervision-based dispute resolution data processing method, the inputting the depth reference feature map into a second spatial attention module of the second convolutional neural network to obtain a second spatial attention map includes: convolution encoding the depth reference feature map using convolution layers of the second spatial attention module to obtain a second spatial attention score map; and non-linearly activating the second spatial attention score map using a Softmax activation function to obtain the second spatial attention score map.
In the dispute resolution data processing method based on credit supervision, the calculating a difference feature map between the detection feature map and the reference feature map includes: calculating a difference characteristic diagram between the detection characteristic diagram and the reference characteristic diagram according to the following formula;
wherein the formula is:
wherein
Is the detection characteristic diagram of the image,
is the reference characteristic diagram that is described,
is the detection characteristic map
Is determined for each of the characteristic values of (a),
is the detection characteristic map
A global mean of all feature values of, and
is the detection characteristic diagram
The size of (a) is greater than (b),
is a weighted hyperparameter.
In the dispute resolution data processing method based on credit supervision, the passing the difference feature map through a classifier to obtain a classification result includes: processing the differential feature map to generate a classification result according to the following formula; wherein the formula is:
In the dispute mediation data processing method based on credit supervision, the evidence material is a credit commitment or a mediation agreement.
According to another aspect of the present application, there is provided a credit supervision-based dispute resolution data processing system, comprising:
the material uploading module is used for acquiring evidence materials to be uploaded to the block chain network;
the user signature intercepting module is used for intercepting a user signature area from the evidence material;
a super-pixel user signature area image generation module, configured to process the user signature area using a SLIC algorithm to generate a super-pixel user signature area image;
the reference signature image acquisition module is used for acquiring a reference signature image;
the feature extraction module is used for enabling the super-pixel user signature area image and the reference signature image to pass through a twin network comprising a first convolutional neural network and a second convolutional neural network to obtain a detection feature map and a reference feature map, wherein the first convolutional neural network and the second convolutional neural network have the same network structure;
a difference module for calculating a difference feature map between the detection feature map and the reference feature map; and
and the classification module is used for enabling the differential feature map to pass through a classifier to obtain a classification result, and the classification result is used for indicating whether the signature in the evidence material is real or not.
According to still another aspect of the present application, there is provided an electronic apparatus including: a processor; and a memory having stored therein computer program instructions which, when executed by the processor, cause the processor to perform the credit supervision-based dispute resolution data processing method as described above.
According to yet another aspect of the present application, there is provided a computer readable medium having stored thereon computer program instructions which, when executed by a processor, cause the processor to perform the credit supervision-based dispute resolution data processing method as described above.
Compared with the prior art, the dispute mediation data processing method and system based on credit supervision provided by the application can be used for intelligently checking the evidence material to be uploaded through comparison between the signature in the evidence material and the reference signature so as to ensure the authenticity of the evidence material. And then improve the accuracy of carrying out the intelligent verification to the evidence material of treating to upload in order to ensure its authenticity of uploading the data.
Drawings
The above and other objects, features and advantages of the present application will become more apparent by describing in more detail embodiments of the present application with reference to the attached drawings. The accompanying drawings are included to provide a further understanding of the embodiments of the application and are incorporated in and constitute a part of this specification, illustrate embodiments of the application and together with the description serve to explain the principles of the application. In the drawings, like reference numbers generally represent like parts or steps.
Fig. 1 illustrates a functional block diagram of a credit supervision dispute system according to an embodiment of the present application.
FIG. 2 illustrates a credit mediation business flow diagram according to an embodiment of the application.
FIG. 3 is a diagram illustrating a credit supervision dispute resolution system architecture according to an embodiment of the present application
FIG. 4 illustrates a credit commitment template according to an embodiment of the application.
Fig. 5 is a flowchart illustrating a dispute mediation data processing method based on credit supervision according to an embodiment of the present application.
Fig. 6 illustrates an architecture diagram of a dispute mediation data processing method based on credit supervision according to an embodiment of the present application.
Fig. 7 is a flowchart illustrating a first convolutional neural network encoding process in a credit supervision-based dispute resolution data processing method according to an embodiment of the present application.
Fig. 8 illustrates a flowchart of a first spatial attention map generation process in a dispute resolution data processing method based on credit supervision according to an embodiment of the present application.
FIG. 9 illustrates a block diagram of a credit oversight-based dispute resolution data processing system according to an embodiment of the present application.
Detailed Description
Hereinafter, example embodiments according to the present application will be described in detail with reference to the accompanying drawings. It should be understood that the described embodiments are only some embodiments of the present application and not all embodiments of the present application, and that the present application is not limited by the example embodiments described herein.
Overview of scenes
As described above, before a proof material such as "letter of promise for credit" and "mediation agreement" is linked to a credible proof, although a block link network can guarantee that the proof material is non-tamper-proof, it cannot guarantee the authenticity of the proof before linking, for example, a problem of a counterfeit signature occurs. Therefore, in order to improve the effectiveness of dispute mediation based on credit supervision, it is expected that the evidence material to be uploaded is intelligently verified to ensure the authenticity of the uploaded data.
In the technical scheme of the application, the uploaded evidence material can be intelligently checked through comparison between the signature in the evidence material and the reference signature so as to ensure the authenticity of the evidence material. Specifically, firstly, evidence material to be uploaded to the blockchain network is obtained, and then a user signature area is intercepted from the evidence material. For example, the user signature region may be intelligently intercepted from the evidence material through a target detection network, or may be intercepted from the evidence material by a manual interception method, which is not limited in this application.
Considering that the evidence material is usually a scanned material, the image definition itself is not high, so that the definition of the user signature region obtained by image capture is difficult to guarantee. In order to ensure the contrast accuracy, in the technical scheme of the application, the user signature region is processed by using a SLIC algorithm to generate a super-pixel user signature region image. Here, the super-pixel user signature area image is generated by calculating the color mean value of each cluster after performing SLIC segmentation processing on the original image of the user signature area, and it is essential that the pixels with similar colors and similar positions are subjected to micro-merging in advance. Because the subsequent region growing algorithm only uses the color features and the space features, effective information still exists in the feature map, and meanwhile, because the super-pixel user signature region image realizes the dimension reduction expression of the original image, the processing speed of the subsequent algorithm is improved, and the identification of the user signature content is facilitated.
Then, a reference signature image is obtained, and in a specific example of the application, the signature image of the user in other files can be crawled as the reference image through a web crawler technology.
After the super-pixel user signature area image and the reference image are obtained, a twin network comprising a first convolutional neural network and a second convolutional neural network is utilized to extract difference feature representation of the super-pixel user signature area image and the reference image in a high-dimensional feature space. Specifically, the first convolutional neural network and the second convolutional neural network of the twin network are used for respectively encoding the superpixel user signature region image and the reference image to obtain a detection feature map and a reference feature map, and then a difference feature map between the detection feature map and the reference feature map is calculated to represent the difference feature representation of the superpixel user signature region image and the reference image in a high-dimensional feature space.
In particular, in the technical solution of the present application, considering that the signatures in the superpixel user signature region image and the reference image have a personal writing style, which is presented on the spatial arrangement style of the user signature, the first convolutional neural network and the second convolutional neural network with a spatial attention mechanism are used to capture the identifiable features of the user signature on the spatial distribution, and give higher weight to the identifiable features so as to make the identifiable features more prominent in the final detected feature map and the reference feature map.
In the technical scheme of the application, the SLIC algorithm is considered to be based on pixel clustering to express neighborhood features, and the neighborhood feature extraction effect similar to the convolution kernel of the convolution neural network is achieved. Therefore, when the super-pixel user signature region image and the reference signature image pass through a twin network including a first convolutional neural network and a second convolutional neural network to obtain a detection feature map and a reference feature map, the detection feature map may be considered to have a deeper layer depth relative to the reference feature map. Therefore, if the differential feature map between the detection feature map and the reference feature map is directly calculated, the classification effect of the differential feature map may be affected due to the difference in layer depth between the two.
Therefore, preferably, a mode of attention-oriented hierarchical depth simultaneous difference calculation is adopted for the detection feature map and the reference feature map, and specifically includes:
wherein
Is the detection characteristic diagram of the image,
is the reference characteristic diagram that is the graph,
is the detection characteristic diagram
Is determined by the characteristic values of the first and second characteristic values,
is the detection characteristic map
A global mean of all feature values of, and
is the detection characteristic map
The size of (a) is greater than (b),
is a weighted hyperparameter.
That is, the detected feature map as a deep feature
As an attention-guiding weight to the reference feature map as a shallow feature
Applying a consistent attention mechanism of sub-dimension distribution to perform volume matching between high-dimensional manifolds with depth difference, thereby enabling the detection feature map
And the reference characteristic diagram
Simultaneous distribution with high consistency in each sub-dimension, thereby performing differential feature mapping
Can suppress the detection signature
And the reference characteristic diagram
The difference in layer depth of the layer affects the differential feature map
To improve the differential feature map
The classification effect of (1). Therefore, the accuracy of intelligent verification of the evidence materials to be uploaded is improved, and the authenticity of the uploaded data is ensured.
Based on this, the application provides a dispute mediation data processing method based on credit supervision, which includes: acquiring evidence materials to be uploaded to a block chain network; intercepting a user signature region from the evidence material; processing the user signature region using a SLIC algorithm to generate a superpixel user signature region image; acquiring a reference signature image; passing the super-pixel user signature area image and the reference signature image through a twin network comprising a first convolutional neural network and a second convolutional neural network to obtain a detection feature map and a reference feature map, wherein the first convolutional neural network and the second convolutional neural network have the same network structure; calculating a difference feature map between the detection feature map and the reference feature map; and passing the differential feature map through a classifier to obtain a classification result, wherein the classification result is used for indicating whether a signature in the evidence material is real or not.
Having described the general principles of the present application, various non-limiting embodiments of the present application will now be described with reference to the accompanying drawings.
Exemplary analysis
At present, voluntary principles are adopted in the mediation process of litigation mediation, people mediation, mediation and the like, and due to the fact that some parties lack integrity, lack of supervision on performance, and do not have corresponding penalty measures on duty without commitment, the automatic fulfillment rate is low and the complaint return rate is high after all parties of the parties reach mediation agreements. This not only wastes a lot of judicial resources, but also reduces the judicial credibility of the mediation work. At present, dispute mediators in dispute mediation work mainly enter dispute information and evidence materials through paper file management or a dispute mediation management system adopting a centralized storage mode, dispute file data and electronic data are easily lost and tampered in the mode, and data in the system cannot be used as electronic evidence.
Therefore, by building a set of block chain technology-based digital credit mediation system, besides the basic dispute mediation service function, the system also integrates the functional blocks of credit acceptance, credit supervision, credit information collection, credit restoration and the like to form a closed-loop mediation service, and meanwhile, the evidence materials linked up in the system can be used as credible electronic evidence to be provided for the judicial department. On one hand, the automatic fulfillment rate of the mediation protocol can be improved, and the advantages of the mediation system are guaranteed; disputes are resolved in the mediation stage, and the working pressure of judicial institutions is relieved. And secondly, the mediation cost can be greatly reduced, and the mediation efficiency can be improved. In addition, the credit information generated in the mediation work is collected into a local credit information sharing platform and a judicial institution, so that the credit information contributes to the construction of a social credit system. As shown in fig. 1, it is roughly divided into four modules:
1. establishing a credit mediation mode: a novel credit-based supervision mechanism is introduced in the dispute mediation process, namely a 'prior credit commitment', 'business affairs and credit supervision' and a 'late loss and punishment' working mechanism are implemented in the mediation process.
2. The credit supervision agency participates in mediation: and applying for credit supervision on the distrusted party who does not fulfill the mediation protocol, and assigning a public trust lawyer to perform propaganda and education of citizen integrity and distrustment and punishment by a third-party credit supervision mechanism to urge the public trust lawyer to fulfill the mediation protocol and improve the automatic fulfillment rate of the mediation protocol.
3. Non-public credit information aggregation: the information of the distressed credit parties still not fulfilling the mediation agreement after the credit supervision is collected to judicial institutions, local public credit information platforms, credit investigation institutions and the like, and government departments at all places can carry out distressed credit joint punishment according to local credit regulations, so that 'one districts lose credit, everywhere is limited', and the districts are promoted to fulfill the agreement.
4. Block chain trusted evidence storage technology: dispute information and a mediation process are digitally managed by adopting an informatization technology, and evidential materials such as 'credit commitment', 'mediation agreement' and the like signed by parties in the mediation process are stored in a chain manner, so that traceability of the mediation whole process, non-falsification of the evidential materials and safety compliance application of the dispute mediation information are realized.
Specifically, as shown in fig. 2, a credit mediation system based on a block chain trusted deposit evidence technology is built, mediation services and a credit supervision mechanism are integrated together, and the work quality and efficiency of existing mediation services are improved.
1. An acceptance stage: the moderator asks parties to sign a letter of promise for credit before accepting dispute mediation, and uploads the letter of promise to a digital mediation system (a credit supervision dispute resolving system, as shown in fig. 3) after signing is finished, and then the letter of promise is made by the moderator.
2. And (3) a mediation stage: after the dispute is accepted by the moderator, the basic information of the case is input into the credit supervision dispute resolving system, and the party is required to sign a credit acceptance book (as shown in fig. 4), then the party of the moderator achieves the dispute and signs a mediation agreement book, and the mediation agreement book is uploaded to a digital mediation system (the credit supervision dispute resolving system) and is subjected to block chain trusted deposit.
3. And a fulfillment phase: if the mediation agreement is not fulfilled automatically after the mediation agreement is fulfilled by the party, the mediator may apply for credit supervision of the distrusted party to a third-party credit supervision agency.
4. And (3) a supervision stage: after the third-party credit supervision organization receives the credit supervision application, a public credit lawyer is assigned to intervene investigation, if the fact is verified, the third-party credit supervision organization informs that the reason is not fulfilled or the result that the party not fulfilling the mediation protocol is refused to lose credit through a telephone or a short message, brings the party losing credit into a grey list, automatically transfers the party not fulfilling the obligation to a black list after 7 days, meanwhile, collects the information of losing credit to a judicial organization and a local public credit information platform, realizes credit information sharing, and carries out credit punishment on the party losing credit by local related departments according to social credit rules.
5. And (3) credit restoration stage: after the party of the lost credit takes measures such as fulfilling the mediation agreement and the like, the party of the lost credit submits the repair materials to the system according to local requirements to carry out credit repair, and relevant lost credit records and credit measures are cancelled after the relevant departments confirm that the conditions are met.
Specifically, the operation flow is divided into the following steps:
the first step of system login: the user opens the PC browser and logs in the 'credit supervision dispute resolution system'.
Second step case registration: a user enters a system home page, clicks a 'case management' column, clicks a 'newly-built case' button, inputs case information, and clicks a 'submit' button to store the case information.
And thirdly, uploading evidence: the user selects the case record, clicks the 'evidence uploading' button, selects the corresponding evidence type, clicks the 'add evidence' button, selects the corresponding evidence file, and clicks the 'confirm' button to finish uploading.
Step four, case handling: a user selects a case record, clicks a 'accept case' button, selects a moderator and enters a case moderation stage; clicking the "do not accept case" button, the case does not accept termination mediation.
Fifthly, case mediation: and clicking a 'case mediation' button by the user, clicking a 'mediation record', displaying the record, and clicking 'adding the mediation record' to fill in the mediation record information.
And sixthly, case setting: the user selects the case record, clicks the 'conclusion report' button, inputs the mediation conclusion information, selects 'yes' to upload the promise of credit 'mediation protocol' to mediate, and selects 'no' to terminate the case which is not successfully mediated.
Seventh step method confirmation: and the user mediates and finishes the operation, clicks a 'judicial confirmation' button and selects 'judicial confirmation letter' issued by the court to upload and store the evidence if all the parties need judicial confirmation.
Eighth step, credit supervision: when the distressing party does not fulfill the mediation protocol, the user selects the case record, clicks the 'credit supervision' button, then clicks the 'raise credit supervision' button, selects the party to input the distressing behavior information, and after the 'submission', the system pushes the case information to the third-party credit supervision mechanism to carry out credit supervision propaganda and education activities on the party.
Ninth step credit punishment: the credit is still not fulfilled after the credit supervision, the system brings the party concerned with the loss of credit into a grey list of the loss of credit, the party concerned with the loss of credit who still does not fulfill obligations after 7 days is automatically transferred to a black list, and meanwhile, the information about the loss of credit is collected to a judicial department and a local public credit information platform, so that the credit information sharing is realized, and the local relevant department carries out credit punishment on the party concerned with the loss of credit according to social credit rules.
Tenth step, case quick cutting: when the case can be quickly cut in the court, the user selects the corresponding case, clicks the 'case element table' button, enters the case element information for storage, and then clicks the 'lifting quick cut' button to push the case to the court quick cut system.
Case arbitration in the eleventh step: when a case can be arbitrated in the arbitration mechanism, a user selects the corresponding case, clicks a 'case element table' button, enters case element information for storage, and then clicks a 'lifting arbitration' button to push the case to the arbitration mechanism arbitration system.
Particularly, according to the technical scheme of the application, the automatic fulfillment rate of dispute resolution agreements is improved from 30-40% to 80-90% in the past according to the pilot work carried out by the resolution organization, and the automatic fulfillment rate is greatly improved. Meanwhile, the case processing efficiency is greatly improved, and the overall processing quantity of trial-and-error mediation organization disputes is doubled compared with the same period in the last year.
In summary, the credit is embedded into the whole process of mediation, false mediation cases are filtered through 'prior credit commitment' to improve the case quality, parties are supervised to fulfill the mediation agreement through 'credit supervision', and distrusted to fulfill obligations of distrusted parties through 'distrusted constraint mechanism', so that the honesty awareness and conscious fulfillment awareness of citizens are enhanced, the situations of malicious mediation, false mediation and the like are filtered, the automatic fulfillment rate of cases is improved, judicial resources and mediation resources are saved, the governing capability of the basic level society is improved, and the goal of maintaining harmony and stability of the society is finally achieved.
Exemplary method
Fig. 5 is a flowchart illustrating a dispute mediation data processing method based on credit supervision according to an embodiment of the present application. As shown in fig. 5, a dispute mediation data processing method based on credit supervision according to an embodiment of the present application includes: s110, obtaining evidence materials to be uploaded to a block chain network; s120, intercepting a user signature area from the evidence material; s130, processing the user signature area by using a SLIC algorithm to generate a super-pixel user signature area image; s140, acquiring a reference signature image; s150, enabling the super-pixel user signature area image and the reference signature image to pass through a twin network comprising a first convolutional neural network and a second convolutional neural network to obtain a detection feature map and a reference feature map, wherein the first convolutional neural network and the second convolutional neural network have the same network structure; s160, calculating a difference characteristic diagram between the detection characteristic diagram and the reference characteristic diagram; and S170, passing the differential feature map through a classifier to obtain a classification result, wherein the classification result is used for indicating whether the signature in the evidence material is real or not.
Fig. 6 illustrates an architecture diagram of a dispute mediation data processing method based on credit supervision according to an embodiment of the present application. As shown in fig. 6, in the network structure, first, evidence material to be uploaded to the blockchain network is acquired; then, intercepting a user signature area from the obtained evidence material; processing the user signature area by using an SLIC algorithm to generate a super-pixel user signature area image; acquiring a reference signature image; secondly, enabling the super-pixel user signature area image and the reference signature image to pass through a twin network comprising a first convolutional neural network and a second convolutional neural network to obtain a detection feature map and a reference feature map, wherein the first convolutional neural network and the second convolutional neural network have the same network structure; calculating a difference feature map between the detection feature map and the reference feature map; and then, the differential feature map is processed by a classifier to obtain a classification result, and the classification result is used for indicating whether the signature in the evidence material is real or not.
More specifically, in step S110 and step S120, evidentiary material to be uploaded to the blockchain network is acquired, and a user signature region is intercepted from the evidentiary material. In one specific example of the present application, the evidentiary material is a credit commitment or a mediation agreement.
More specifically, in step S130, the user signature region is processed using the SLIC algorithm to generate a superpixel user signature region image. Considering that the evidence material is usually a scanned material, the image definition itself is not high, so that the definition of the user signature region obtained by image capture is difficult to guarantee. In order to ensure the contrast accuracy, in the technical scheme of the application, the user signature region is processed by using a SLIC algorithm to generate a super-pixel user signature region image. Here, the super-pixel user signature area image is generated by calculating a color mean value of each cluster after SLIC segmentation processing is performed on the original image of the user signature area, and it is essential that pixels with similar colors and similar positions are subjected to micro-merging in advance. Because the subsequent region growing algorithm only uses the color features and the space features, effective information still exists in the feature map, and meanwhile, because the super-pixel user signature region image realizes the dimension reduction expression of the original image, the processing speed of the subsequent algorithm is improved, and the identification of the user signature content is facilitated.
More specifically, in step S140, a reference signature image is acquired. In one specific example of the present application, the reference image may be crawled by web crawler technology with the user's signature image on other documents.
More specifically, in step S150 and step S160, the superpixel user signature area image and the reference signature image are passed through a twin network comprising a first convolutional neural network and a second convolutional neural network to obtain a detection feature map and a reference feature map, wherein the first convolutional neural network and the second convolutional neural network have the same network structure; and then calculating a difference characteristic diagram between the detection characteristic diagram and the reference characteristic diagram. In the technical scheme of the application, a twin network comprising a first convolutional neural network and a second convolutional neural network can be utilized to extract difference feature representations of the superpixel user signature area image and the reference image in a high-dimensional feature space. Specifically, the first convolutional neural network and the second convolutional neural network of the twin network are used for respectively coding the super-pixel user signature region image and the reference image to obtain a detection feature map and a reference feature map, and then a differential feature map between the detection feature map and the reference feature map is calculated to represent the differential feature representation of the super-pixel user signature region image and the reference image in a high-dimensional feature space. In particular, in the technical solution of the present application, considering that the signatures in the superpixel user signature region image and the reference image have a personal writing style, which is presented on the spatial arrangement style of the user signature, the first convolutional neural network and the second convolutional neural network with a spatial attention mechanism are used to capture the identifiable features of the user signature on the spatial distribution, and give higher weight to the identifiable features so as to make the identifiable features more prominent in the final detected feature map and the reference feature map.
Fig. 7 is a flowchart illustrating a first convolutional neural network encoding process in a credit supervision-based dispute resolution data processing method according to an embodiment of the present application. As shown in fig. 7, the first convolutional neural network coding process includes: s210, performing depth convolution coding on the super-pixel user signature area image by using the multiple layers of convolution layers of the first convolution neural network so as to output a depth detection feature map by the last layer of the multiple layers of convolution layers; s220, inputting the depth detection feature map into a first spatial attention module of the first convolutional neural network to obtain a first spatial attention map; and S230, multiplying the position-based points of the depth detection feature map and the first spatial attention map to obtain the detection feature map.
Fig. 8 illustrates a flowchart of a first spatial attention map generation process in a dispute resolution data processing method based on credit supervision according to an embodiment of the present application. As shown in fig. 8, the first spatial attention map generation process includes: s310, carrying out convolution coding on the depth detection feature map by using a convolution layer of the first spatial attention module to obtain a first spatial attention score map; and S320, performing nonlinear activation on the first spatial attention score map using a Softmax activation function to obtain the first spatial attention score map.
In particular, in the technical solution of the present application, it is considered that the SLIC algorithm expresses the neighborhood features substantially based on pixel clustering, and has a similar neighborhood feature extraction effect as the convolution kernel of the convolutional neural network. Therefore, when the super-pixel user signature region image and the reference signature image pass through a twin network including a first convolutional neural network and a second convolutional neural network to obtain a detection feature map and a reference feature map, the detection feature map may be considered to have a deeper layer depth relative to the reference feature map. Therefore, if the differential feature map between the detected feature map and the reference feature map is directly calculated, the classification effect of the differential feature map may be affected due to the difference in layer depth between the two.
Therefore, preferably, a mode of attention-oriented hierarchical depth simultaneous difference calculation is adopted for the detection feature map and the reference feature map, specifically:
wherein
Is the detection characteristic diagram of the image,
is the reference characteristic diagram that is described,
is the detection characteristic diagram
Is determined by the characteristic values of the first and second characteristic values,
is the detection characteristic diagram
A global mean of all feature values of, and
is the detection characteristic diagram
The size of (a) is greater than (b),
is a weighted hyperparameter.
That is, the detected feature map as a deep feature
As attention-guiding weight, to the reference feature map as a shallow feature
Applying a consistent attention mechanism of sub-dimension distribution to perform volume matching between high-dimensional manifolds with depth difference, thereby enabling the detection feature map
And said reference featureDrawing (A)
Simultaneous distribution with high consistency in each sub-dimension, thereby performing differential feature mapping
Can suppress the detection signature
And the reference characteristic diagram
The difference in layer depth of the layer affects the differential feature map
To improve the differential feature map
The classification effect of (1). Therefore, the accuracy of intelligent verification of the evidence materials to be uploaded is improved, and the authenticity of the uploaded data is ensured.
More specifically, in step S170, the differential feature map is passed through a classifier to obtain a classification result, and the classification result is used to indicate whether the signature in the evidence material is authentic.
In summary, a dispute mediation data processing method based on credit supervision is clarified, and the evidence material to be uploaded is intelligently checked through comparison between the signature in the evidence material and the reference signature to ensure authenticity of the evidence material. And then improve the accuracy of carrying out the intelligent verification to the evidence material of treating to upload in order to ensure its authenticity of uploading the data.
Exemplary System
FIG. 9 illustrates a block diagram of a credit oversight-based dispute resolution data processing system according to an embodiment of the present application. As shown in fig. 9, a dispute mediation data processing system 300 based on credit supervision according to the embodiment of the present application includes: a material upload module 310; a superpixel user signature region image generation module 320; a reference signature image acquisition module 330; a feature extraction module 340; a difference module 350; and, a classification module 360.
The material uploading module 310 is configured to obtain an evidence material to be uploaded to a block chain network; the user signature intercepting module 320 is configured to intercept a user signature region from the evidence material; the super-pixel user signature region image generating module 330 is configured to process the user signature region by using an SLIC algorithm to generate a super-pixel user signature region image; the reference signature image obtaining module 340 is configured to obtain a reference signature image; the feature extraction module 350 is configured to pass the super-pixel user signature area image and the reference signature image through a twin network including a first convolutional neural network and a second convolutional neural network to obtain a detected feature map and a reference feature map, where the first convolutional neural network and the second convolutional neural network have the same network structure; the difference module 360 is configured to calculate a difference feature map between the detection feature map and the reference feature map; and the classification module 370 is configured to pass the differential feature map through a classifier to obtain a classification result, where the classification result is used to indicate whether a signature in the evidence material is authentic.
In one example, in the credit supervision-based dispute resolution data processing system 300 described above, the feature extraction module 350 includes: depth convolution encoding the super-pixel user signature area image using a plurality of convolutional layers of the first convolutional neural network to output a depth detection feature map from a last layer of the plurality of convolutional layers; inputting the depth detection feature map into a first spatial attention module of the first convolutional neural network to obtain a first spatial attention map; and calculating the position-based multiplication of the depth detection feature map and the first spatial attention map to obtain the detection feature map.
In one example, in the above-described credit oversight-based dispute resolution data processing system 300, the difference module 360 comprises: calculating a difference feature map between the detection feature map and the reference feature map in the following formula;
wherein the formula is:
wherein
Is the detection characteristic diagram of the image,
is the reference characteristic diagram that is described,
is the detection characteristic diagram
Is determined for each of the characteristic values of (a),
is the detection characteristic map
A global mean of all feature values of, and
is the detection characteristic map
The size of (a) is greater than (b),
is a weighted hyperparameter.
In one example, in the above-described credit oversight-based dispute resolution data processing system 300, the classification module 370 comprises: processing the differential feature map to generate a classification result according to the following formula; wherein the formula is:
In summary, a dispute resolution data processing system based on credit supervision according to the embodiment of the present application is set forth, which intelligently verifies the evidence material to be uploaded to ensure the authenticity of the evidence material through comparison between the signature in the evidence material and the reference signature. And then the accuracy of intelligent verification of the evidence materials to be uploaded is improved so as to ensure the authenticity of the uploaded data.
Claims (10)
1. A dispute mediation data processing method based on credit supervision is characterized by comprising the following steps:
acquiring evidence materials to be uploaded to a block chain network;
intercepting a user signature region from the evidence material;
processing the user signature region using a SLIC algorithm to generate a superpixel user signature region image;
acquiring a reference signature image;
passing the super-pixel user signature area image and the reference signature image through a twin network comprising a first convolutional neural network and a second convolutional neural network to obtain a detection feature map and a reference feature map, wherein the first convolutional neural network and the second convolutional neural network have the same network structure;
calculating a difference feature map between the detection feature map and the reference feature map; and
and passing the differential feature map through a classifier to obtain a classification result, wherein the classification result is used for indicating whether the signature in the evidence material is real or not.
2. The credit supervision-based dispute mediation data processing method according to claim 1, wherein the passing the superpixel user signature region image and the reference signature image through a twin network comprising a first convolutional neural network and a second convolutional neural network to obtain a detection feature map and a reference feature map comprises:
depth convolution encoding the super-pixel user signature area image using a plurality of convolutional layers of the first convolutional neural network to output a depth detection feature map from a last layer of the plurality of convolutional layers;
inputting the depth detection feature map into a first spatial attention module of the first convolutional neural network to obtain a first spatial attention map; and
and calculating the position-based multiplication of the depth detection feature map and the first spatial attention map to obtain the detection feature map.
3. The credit supervision-based dispute resolution data processing method according to claim 2, wherein the inputting the depth detection feature map into a first spatial attention module of the first convolutional neural network to obtain a first spatial attention map comprises:
convolution coding the depth detection feature map using convolution layers of the first spatial attention module to obtain a first spatial attention score map; and
nonlinearly activating the first spatial attention score map using a Softmax activation function to obtain the first spatial attention score map.
4. The credit supervision-based dispute resolution data processing method according to claim 3, wherein the passing the superpixel user signature area image and the reference signature image through a twin network comprising a first convolutional neural network and a second convolutional neural network to obtain a detection feature map and a reference feature map comprises:
depth convolution encoding the superpixel user signature region image using a plurality of convolutional layers of the second convolutional neural network to output a depth reference feature map from a last layer of the plurality of convolutional layers;
inputting the depth reference feature map into a second spatial attention module of the second convolutional neural network to obtain a second spatial attention map; and
and calculating the position-based multiplication of the depth reference feature map and the second spatial attention map to obtain the reference feature map.
5. The credit supervision-based dispute resolution data processing method according to claim 4, wherein the inputting the depth reference feature map into a second spatial attention module of the second convolutional neural network to obtain a second spatial attention map comprises:
convolution encoding the depth reference feature map using convolution layers of the second spatial attention module to obtain a second spatial attention score map; and
nonlinearly activating the second spatial attention score map using a Softmax activation function to obtain the second spatial attention score map.
6. The credit supervision-based dispute mediation data processing method according to claim 5, wherein the calculating of the difference feature map between the detection feature map and the reference feature map comprises: calculating a difference feature map between the detection feature map and the reference feature map in the following formula;
wherein the formula is:
wherein
Is the detection characteristic diagram of the image,
is the reference characteristic diagram that is described,
is the detection characteristic diagram
Is determined for each of the characteristic values of (a),
is the detection characteristic map
Is a global mean of all feature values of, and
is the detection characteristic diagram
The size of (a) is greater than (b),
is a weighted hyperparameter.
7. The credit supervision-based dispute resolution data processing method according to claim 6, wherein the passing the differential feature map through a classifier to obtain a classification result comprises: processing the differential feature map using the classifier in the following formula to generate a classification result;
wherein the formula is:
in which
Representing the projection of the difference feature map as a vector,
to
Is a weight matrix of the fully connected layers of each layer,
to
A bias matrix representing the layers of the fully connected layer.
8. A credit supervision based dispute resolution data processing method according to claim 7, wherein the evidence material is a credit commitment or a resolution agreement.
9. A dispute resolution data processing system based on credit supervision is characterized by comprising:
the material uploading module is used for acquiring evidence materials to be uploaded to the block chain network;
the user signature intercepting module is used for intercepting a user signature area from the evidence material;
a super-pixel user signature area image generation module, configured to process the user signature area using a SLIC algorithm to generate a super-pixel user signature area image;
the reference signature image acquisition module is used for acquiring a reference signature image;
the feature extraction module is used for enabling the super-pixel user signature area image and the reference signature image to pass through a twin network comprising a first convolutional neural network and a second convolutional neural network to obtain a detection feature map and a reference feature map, wherein the first convolutional neural network and the second convolutional neural network have the same network structure;
a difference module for calculating a difference feature map between the detection feature map and the reference feature map; and
and the classification module is used for enabling the differential feature map to pass through a classifier to obtain a classification result, and the classification result is used for indicating whether the signature in the evidence material is real or not.
10. The credit supervision-based dispute resolution data processing system of claim 9, wherein the differencing module is further to: processing the differential feature map using the classifier in the following formula to generate a classification result;
wherein the formula is:
wherein
Representing the projection of the difference feature map as a vector,
to
Is a weight matrix of the fully connected layers of each layer,
to
A bias matrix representing the layers of the fully connected layer.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202211326958.0A CN115393966B (en) | 2022-10-27 | 2022-10-27 | Dispute mediation data processing method and system based on credit supervision |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202211326958.0A CN115393966B (en) | 2022-10-27 | 2022-10-27 | Dispute mediation data processing method and system based on credit supervision |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN115393966A true CN115393966A (en) | 2022-11-25 |
| CN115393966B CN115393966B (en) | 2023-01-10 |
Family
ID=84129460
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202211326958.0A Active CN115393966B (en) | 2022-10-27 | 2022-10-27 | Dispute mediation data processing method and system based on credit supervision |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN115393966B (en) |
Citations (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US9436895B1 (en) * | 2015-04-03 | 2016-09-06 | Mitsubishi Electric Research Laboratories, Inc. | Method for determining similarity of objects represented in images |
| CN110378224A (en) * | 2019-06-14 | 2019-10-25 | 香港理工大学深圳研究院 | A kind of detection method of feature changes, detection system and terminal |
| CN114360071A (en) * | 2022-01-11 | 2022-04-15 | 北京邮电大学 | Method for realizing off-line handwritten signature verification based on artificial intelligence |
| CN114898473A (en) * | 2022-05-09 | 2022-08-12 | 中国建设银行股份有限公司 | Handwritten signature comparison method and device and electronic equipment |
| CN114898472A (en) * | 2022-04-26 | 2022-08-12 | 华南理工大学 | Signature identification method and system based on twin vision Transformer network |
| CN114926746A (en) * | 2022-05-25 | 2022-08-19 | 西北工业大学 | SAR image change detection method based on multi-scale differential feature attention mechanism |
-
2022
- 2022-10-27 CN CN202211326958.0A patent/CN115393966B/en active Active
Patent Citations (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US9436895B1 (en) * | 2015-04-03 | 2016-09-06 | Mitsubishi Electric Research Laboratories, Inc. | Method for determining similarity of objects represented in images |
| CN110378224A (en) * | 2019-06-14 | 2019-10-25 | 香港理工大学深圳研究院 | A kind of detection method of feature changes, detection system and terminal |
| CN114360071A (en) * | 2022-01-11 | 2022-04-15 | 北京邮电大学 | Method for realizing off-line handwritten signature verification based on artificial intelligence |
| CN114898472A (en) * | 2022-04-26 | 2022-08-12 | 华南理工大学 | Signature identification method and system based on twin vision Transformer network |
| CN114898473A (en) * | 2022-05-09 | 2022-08-12 | 中国建设银行股份有限公司 | Handwritten signature comparison method and device and electronic equipment |
| CN114926746A (en) * | 2022-05-25 | 2022-08-19 | 西北工业大学 | SAR image change detection method based on multi-scale differential feature attention mechanism |
Non-Patent Citations (3)
| Title |
|---|
| KIAN AHRABIAN ETC.: "On Usage of Autoencoders and Siamese Networks for Online Handwritten Signature Verication", 《ARXIV:1712.02781V2》 * |
| 赵景晨: "基于超像素和孪生卷积神经网络的无监督高分辨率多光谱遥感影像变化检测技术", 《中国优秀硕士学位论文全文数据库 信息科技辑》 * |
| 陈志旺等: "基于目标感知特征筛选的孪生网络跟踪算法", 《光学学报》 * |
Also Published As
| Publication number | Publication date |
|---|---|
| CN115393966B (en) | 2023-01-10 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| Hsiao et al. | Employing blockchain technology to strengthen security of wireless sensor networks | |
| CN106788987A (en) | A kind of collage-credit data based on block chain is shared and transaction system | |
| Hossain et al. | Vehicle registration and information management using blockchain based distributed ledger from bangladesh perspective | |
| CN113434269A (en) | Block chain-based distributed privacy calculation method and device | |
| CN111143339B (en) | Method, device, equipment and storage medium for distributing service resources | |
| DE112021000608T5 (en) | FASTER VIEW CHANGES FOR A BLOCKCHAIN | |
| CN111860865B (en) | Model construction and analysis method, device, electronic equipment and medium | |
| CN112418520A (en) | Credit card transaction risk prediction method based on federal learning | |
| CN111681091A (en) | Financial risk prediction method and device based on time domain information and storage medium | |
| CN109840751A (en) | A kind of electronic bidding book generates online and signature method and e-bidding system | |
| CN110830259A (en) | Method and system for providing originality and integrity certification for multimedia data | |
| CN110309261A (en) | A kind of electronic bidding book generates online and signature method and e-bidding system | |
| CN112995201B (en) | Resource value evaluation processing method based on cloud platform and related device | |
| DE112021004008T5 (en) | VALIDATION OF TRACKED PORTIONS OF RECEIVED SENSOR DATA USING CRYPTOGRAPHIC COMPUTER PROCESSING | |
| CN114462624A (en) | Method for developing credible federal learning based on block chain | |
| CN115393966B (en) | Dispute mediation data processing method and system based on credit supervision | |
| CN111932365B (en) | Financial credit investigation system and method based on block chain | |
| CN114863430A (en) | Automatic population information error correction method, device and storage medium thereof | |
| Parlak et al. | Tamper-proof evidence via blockchain for autonomous vehicle accident monitoring | |
| CN113688418B (en) | Engineering order settlement method and system based on blockchain data storage | |
| CN105913248B (en) | Online payment system based on mobile internet service application | |
| US20230070625A1 (en) | Graph-based analysis and visualization of digital tokens | |
| CN114358767A (en) | Data transaction flow compliance notarization method and device, electronic equipment and storage medium | |
| CN114139206A (en) | Multi-user heterogeneous data merging and concurrent certification method based on block chain privacy protection | |
| Abunadi | Characteristics of electronic integrated system and trust in the provider of service |
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 | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |