Image auditing method, device and equipment and computer medium

Abstract

The invention provides an image auditing method, device and equipment and a computer readable medium, and the method comprises the steps: carrying out the character detection of a to-be-audited image through calling a character detection model, cutting the to-be-audited image according to the detected characters, obtaining at least one character image, and carrying out the cutting of the to-be-audited image according to each character image, performing multi-dimensional feature detection on the character image, selecting a corresponding special text recognition model to perform text recognition on the character image according to a feature detection result to obtain a text recognition result, and performing violation judgment according to all the text recognition results corresponding to the to-be-audited image to obtain a violation judgment result. According to the method and the device, the to-be-audited image which is judged to be illegal is processed according to the preset illegal image processing rule, so that an image auditing solution which is high in automation degree, high in accuracy and high in processing speed is provided, the burden of manual auditing can be greatly reduced, and the accuracy and the efficiency of content auditing are improved.

Description

Technical Field

[0001] The present application relates to the field of computer technology, and in particular to an image review method, device, equipment and computer-readable medium. Background Art

[0002] This section is intended to provide a background or context to the embodiments of the application that are recited in the claims. The description herein is not admitted to be prior art by inclusion in this section.

[0003] With the rapid development of the Internet, especially the popularity of social media, online forums, and user-generated content (UGC) platforms, the amount of image information has exploded. Massive amounts of image content are uploaded and shared on these platforms every day. The diversity and complexity of image content have brought huge challenges to content management.

[0004] Traditional image content review mainly relies on manual review, which has many shortcomings. First, the speed of manual review is limited by individual reading ability and energy. When faced with massive amounts of data, the review efficiency is low and it is difficult to process in a timely manner. Second, manual review is easily affected by subjective factors, which may lead to missed or misjudgment, affecting user experience and platform response speed. Third, the cost of manual review is high, and it is impossible to achieve a 24-hour uninterrupted working mode, which is difficult to meet the needs of modern society for instant information processing.

[0005] In view of the above problems, there is an urgent need to provide an efficient and accurate image review method. Summary of the invention

[0006] Multiple aspects of the present application provide an image review method, apparatus, device and computer-readable storage medium to improve the accuracy and efficiency of image content review.

[0007] In one aspect of the present application, there is provided an image review method, wherein the method comprises:

[0008] Calling a text detection model to perform text detection on the image to be reviewed, and cropping the image to be reviewed according to the detected text to obtain at least one text image;

[0009] For each of the text images, multi-dimensional feature detection is performed on the text image, and a corresponding dedicated text recognition model is selected according to the feature detection result to perform text recognition on the text image to obtain a text recognition result;

[0010] Performing violation judgment according to all the text recognition results corresponding to the image to be reviewed to obtain a violation judgment result;

[0011] The to-be-audited image judged to be in violation is processed according to preset violation image processing rules.

[0012] Another aspect of the present application provides an image review device, wherein the device comprises:

[0013] A text detection module, used to call a text detection model to perform text detection on the image to be reviewed, and to crop the image to be reviewed according to the detected text to obtain at least one text image;

[0014] A text recognition module is used to perform multi-dimensional feature detection on each of the text images, and select a corresponding dedicated text recognition model to perform text recognition on the text image according to the feature detection result to obtain a text recognition result;

[0015] A violation judgment module, used to make a violation judgment based on all the text recognition results corresponding to the image to be reviewed, and obtain a violation judgment result;

[0016] The illegal image processing module is used to process the to-be-audited image that is judged to be illegal according to a preset illegal image processing rule.

[0017] Another aspect of the present application provides an electronic device, the electronic device comprising:

[0018] at least one processor; and

[0019] a memory communicatively connected to the at least one processor; wherein,

[0020] The memory stores instructions that can be executed by the at least one processor, and the instructions are executed by the at least one processor to enable the at least one processor to perform the image review method as described above.

[0021] In another aspect of the present application, a computer-readable storage medium is provided, on which computer program instructions are stored. The computer program instructions can be executed by a processor to implement the image review method as described above.

[0022] In the solution provided by the embodiment of the present application, a text detection model is called to perform text detection on the image to be reviewed, and the image to be reviewed is cropped according to the detected text to obtain at least one text image, and then for each of the text images, multi-dimensional feature detection is performed on the text image, and the corresponding dedicated text recognition model is selected according to the feature detection result to perform text recognition on the text image to obtain a text recognition result, and then a violation judgment is performed according to all the text recognition results corresponding to the image to be reviewed to obtain a violation judgment result, and then the image to be reviewed judged to be a violation is processed according to the preset violation image processing rules, thereby realizing automated review of the image, which can not only improve the review efficiency and ensure timely processing of a large amount of image review work, but also improve the recognition accuracy by selecting the most appropriate dedicated text recognition model, and reduce recognition errors caused by differences in text formats and the like. Since violation judgment is performed based on all text recognition results, it can also effectively detect and process illegal content to ensure the information security of the platform. In summary, the present application provides an image review solution with high automation, high accuracy and fast processing speed through the above-mentioned implementation method, which can greatly reduce the burden of manual review and improve the overall quality and efficiency of content review. BRIEF DESCRIPTION OF THE DRAWINGS

[0023] In order to more clearly illustrate the technical solutions in the embodiments of the present application, a brief introduction will be given below to the drawings required for use in the embodiments or the description of the prior art. Obviously, the drawings described below are some embodiments of the present application. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying any creative labor.

[0024] Other features, objects and advantages of the present application will become more apparent by reading the detailed description of non-limiting embodiments made with reference to the following drawings:

[0025] Figure 1 A schematic diagram of a flow chart of an image review method provided in one embodiment of the present application;

[0026] Figure 2 A flowchart of a specific example of an image review method provided by an embodiment of the present application;

[0027] Figure 3 A schematic diagram of the structure of an image review device provided in one embodiment of the present application;

[0028] Figure 4 A schematic diagram of the structure of an electronic device suitable for implementing the solution in the embodiments of the present application;

[0029] The same or similar reference numerals in the drawings represent the same or similar components. DETAILED DESCRIPTION

[0030] In order to make the purpose, technical solution and advantages of the embodiments of the present application clearer, the technical solution in the embodiments of the present application will be clearly and completely described below in conjunction with the drawings in the embodiments of the present application. Obviously, the described embodiments are part of the embodiments of the present application, not all of the embodiments. Based on the embodiments in the present application, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of this application.

[0031] In a typical configuration of the present application, the terminal and the equipment of the service network each include one or more processors (CPU), input / output interface, network interface and memory.

[0032] The memory may include non-permanent storage in a computer-readable medium, random access memory (RAM) and / or non-volatile memory in the form of read-only memory (ROM) or flash RAM. The memory is an example of a computer-readable medium.

[0033] Computer readable media include permanent and non-permanent, removable and non-removable media, and can be implemented by any method or technology to store information. Information can be computer program instructions, data structures, modules of programs or other data. Examples of computer storage media include, but are not limited to, phase change memory (PRAM), static random access memory (SRAM), dynamic random access memory (DRAM), other types of random access memory (RAM), read-only memory (ROM), electrically erasable programmable read-only memory (EEPROM), flash memory or other memory technology, read-only compact disk (CD-ROM), digital versatile disk (DVD) or other optical storage, magnetic cassettes, magnetic tape disk storage or other magnetic storage devices or any other non-transmission media that can be used to store information that can be accessed by a computing device.

[0034] In view of the problems of low efficiency and poor accuracy in the prior art of image review that relies on manual methods, the embodiments of the present application provide an image review method, apparatus, device and computer-readable storage medium to improve the accuracy and efficiency of image content review. It performs text detection on the image to be reviewed by calling a text detection model, and crops the image to be reviewed according to the detected text to obtain at least one text image. Then, for each of the text images, multi-dimensional feature detection is performed on the text image, and a corresponding dedicated text recognition model is selected according to the feature detection result to perform text recognition on the text image to obtain a text recognition result. Then, a violation judgment is made according to all the text recognition results corresponding to the image to be reviewed to obtain a The violation judgment result is obtained, and then the image to be reviewed that is judged to be in violation is processed according to the preset violation image processing rules, so as to realize automated review of the image, which can not only improve the review efficiency and ensure the timely processing of a large number of image review work, but also improve the recognition accuracy by selecting the most suitable dedicated text recognition model, and reduce recognition errors caused by text format differences and other reasons. Since the violation judgment is based on all text recognition results, it can also effectively detect and process illegal content to ensure the information security of the platform. In summary, this application provides an image review solution with high degree of automation, strong accuracy and fast processing speed through the above implementation method, which can greatly reduce the burden of manual review and improve the overall quality and efficiency of content review.

[0035] The following is an exemplary description of the image review method, device, equipment and computer-readable storage medium provided in this application.

[0036] In actual scenarios, the execution subject of the image review method provided in the embodiment of the present application can be a user device, or a device formed by integrating a user device and a network device through a network, or an application program (such as a machine review system) running on the above device. The user device includes but is not limited to various terminal devices such as computers, mobile phones, tablet computers, smart watches, and bracelets. The network device includes but is not limited to network hosts, single network servers, multiple network server sets, or cloud computing-based computer collections. Here, the cloud is composed of a large number of hosts or network servers based on cloud computing, where cloud computing is a type of distributed computing, a virtual computer composed of a group of loosely coupled computer sets.

[0037] Figure 1 The following is a processing flow of an image review method provided by an embodiment of the present application, wherein the method at least includes the following processing steps:

[0038] Step S101 , calling a text detection model to perform text detection on an image to be reviewed, and cropping the image to be reviewed according to the detected text to obtain at least one text image.

[0039] Among them, the text detection model refers to an algorithm model trained through machine learning, which can recognize text areas in images.

[0040] In the specific implementation of this step, the image to be reviewed can be first input into the text detection model, which will scan the image, mark the areas in the image that may contain text, and annotate the location information of these areas. Next, the original image is cropped according to the text location information output by the model, so as to extract the image fragment containing text, that is, the text image. By removing irrelevant parts of the image, the amount of data for subsequent processing can be reduced, the processing speed can be improved, and at the same time, the processing focus can be ensured to be on the text part containing important information, thereby improving the accuracy of recognition.

[0041] For example, when processing a street view photo containing a billboard, the text detection model can identify the text on the billboard and mark its bounding box. Then, based on the bounding box information, the algorithm will crop the original image to obtain an image segment containing only the text on the billboard for further feature analysis and text recognition. This method not only ensures that the valuable part of the image is processed, but also avoids non-text areas from interfering with the recognition results, thereby improving the overall review efficiency and accuracy.

[0042] It should be noted that the above-mentioned text detection model can be implemented using any text detection model provided by the prior art. After being trained with a large number of samples, it can be used in the present application to detect and extract text images from the images to be reviewed. The embodiments of the present application do not limit the specific content of the text detection model.

[0043] Step S102: for each of the text images, perform multi-dimensional feature detection on the text image, and select a corresponding dedicated text recognition model to perform text recognition on the text image according to the feature detection result to obtain a text recognition result.

[0044] One of the purposes of this step is to improve the accuracy of recognition by analyzing different features of the text image and selecting the most suitable recognition model. Among them, multi-dimensional feature detection can refer to the detection of multiple features of the text image, including but not limited to rotation, vertical arrangement, long text and language classification.

[0045] When this step is implemented, a series of pre-trained models can be first applied to extract features from the text image. For example, a vertical text detection model can be used to detect whether the text is vertical; a long text detection model can be used to determine whether the text belongs to a long text; and finally, a text language classification model can be used to determine the language type of the text. According to the results of the above feature detection, the system can select the most appropriate special text recognition model to recognize the text image. For example, if the feature detection result shows that the text is vertical, a recognition model that specifically processes vertical text is called to perform text recognition on the text image; if the text is a long text, a recognition model that specifically processes long text is called to perform text recognition on the text image; if the text belongs to a specific language, the corresponding language recognition model is called to perform text recognition on the text image. In this way, not only can the accuracy of text recognition be improved, but also the recognition requirements of different types of text can be adapted, thereby ensuring the effectiveness and reliability of the entire recognition process.

[0046] For example, when processing an image containing vertical Chinese text, the system first uses the vertical text detection model to identify that the text is arranged vertically, and then selects a recognition model designed specifically for vertical Chinese to recognize the text content. This not only avoids the recognition errors that may occur when general recognition models process vertical text, but also significantly improves the quality of recognition results, thus ensuring that the system has higher accuracy and robustness when processing complex and diverse image content.

[0047] Step S103, performing violation judgment according to all the text recognition results corresponding to the image to be reviewed to obtain a violation judgment result.

[0048] One of the purposes of this step is to comprehensively analyze all the identified text content and evaluate whether the content meets the preset violation standards to determine whether the image should be considered as a violation. Violation judgment can refer to the process of evaluating the identified text string based on preset rules, with the purpose of identifying whether the text contains illegal information, such as sensitive words or other text content that does not comply with regulations.

[0049] In the specific implementation of this step, after obtaining all the text recognition results, these results can be combined into a text string and sent to the violation judgment module, wherein, by combining the text string and then making a violation judgment, more comprehensive context information can be provided, which helps to improve the accuracy of the violation judgment. The above-mentioned violation judgment module can include two main components: a violation keyword matching model (also known as a sensitive word matching model, where sensitive words are violation keywords) and a violation text recognition model. The sensitive word matching model can compare the identified text recognition results with a preset sensitive word list. If any sensitive words are included in the text recognition results, the image to be reviewed will be marked as having violation information. On the other hand, the violation text recognition model uses a neural network model to identify whether the text recognition results are illegal or vulgar. If the evaluation result exceeds the preset risk threshold, the image to be reviewed will also be marked as having violation information. The violation text recognition model is usually built based on a neural network, trained by a large amount of annotated text data, and learns to identify possible violations or vulgar content in the text. The model will analyze the content characteristics of the text and output a risk score. If this score exceeds the preset risk threshold, the system can determine that the text content contains violation information and mark the corresponding image to be reviewed. For example, the model may identify sensitive words such as "illegal", "gambling", and "pornography" and score them based on their frequency of occurrence and context. By adopting the illegal text recognition model, the system can automatically evaluate the compliance of the text content, thereby improving the accuracy and efficiency of violation judgments.

[0050] For example, suppose there is a picture containing user comments, and the comment text is identified as "There are some inappropriate remarks here" through optical character recognition (OCR) technology. Then, the system sends the text string to the sensitive word matching model, and the model checks whether the text string contains words in the preset sensitive word list. If the comment contains words in the sensitive word list, the system can immediately mark the image as containing illegal information. In addition, the system also sends the text string to the text review model, which can evaluate whether the content of the text string has a risk of violation. If the evaluation result shows that the risk level exceeds the set threshold, the system can also mark the image as containing illegal information. In this way, not only can the illegal content be identified quickly and accurately, but also the false alarm rate can be effectively reduced to ensure that only images that are truly illegal will be marked and processed. Therefore, this embodiment can significantly improve the accuracy and efficiency of image content review and ensure the security and compliance of platform content.

[0051] Step S104: Process the image to be reviewed that is judged to be in violation according to a preset rule for processing the image in violation.

[0052] One of the purposes of this step is to decide how to deal with images that have been identified as containing illegal content through a pre-set rule system. Among them, the illegal image processing rules refer to a series of guiding principles and specific operating guidelines for handling illegal images formulated in accordance with platform policies and laws and regulations.

[0053] When this step is implemented, after completing the violation judgment of all text strings in the image, specific processing measures can be determined based on the violation judgment results and with reference to the preset violation image processing rules. These measures usually include but are not limited to marking the image as suspected violation and submitting it for manual review or directly refusing to display it. For example, if the sensitive word matching model finds a high-risk sensitive word, the system can directly submit it for manual review, and if the violation risk level assessed by the text review model exceeds a certain threshold, the system can also submit it for manual review. In addition, the system can also record relevant information of the violation image, such as the violation type, processing time, and processing personnel, for later audit and improvement.

[0054] For example, if the text in an image is identified as containing sensitive words, according to the preset illegal image processing rules, the system can immediately mark the image as illegal and automatically trigger further manual review processes. In this process, the system can not only ensure that illegal content will not be directly published without processing, but also support subsequent audit work by recording the specific details of the illegal incident. In this way, not only can the spread of illegal content be effectively controlled, but also the transparency and traceability of the review process can be improved, ensuring the compliance of platform operations and the healthy environment of the user community. Therefore, the accuracy and efficiency of image review can be significantly improved, while ensuring the security and compliance of platform content.

[0055] The above-mentioned image review method provided in the embodiment of the present application performs text detection on the image to be reviewed by calling the text detection model, and crops the image to be reviewed according to the detected text to obtain at least one text image, and then performs multi-dimensional feature detection on each of the text images, and selects the corresponding dedicated text recognition model according to the feature detection result to perform text recognition on the text image to obtain a text recognition result, and then performs violation judgment according to all the text recognition results corresponding to the image to be reviewed to obtain a violation judgment result, and then processes the image to be reviewed judged to be in violation according to the preset violation image processing rules, thereby realizing automated review of the image, which can not only improve the review efficiency and ensure timely processing of a large amount of image review work, but also improve the recognition accuracy by selecting the most appropriate dedicated text recognition model, and reduce recognition errors caused by text format differences and other reasons. Since violation judgment is based on all text recognition results, it can also effectively detect and process illegal content to ensure the information security of the platform. In summary, the present application provides an image review solution with high automation, high accuracy and fast processing speed through the above-mentioned implementation method, which can greatly reduce the burden of manual review and improve the overall quality and efficiency of content review.

[0056] In some modified implementations, the performing multi-dimensional feature detection on the text image includes at least one of the following:

[0057] Calling a pre-trained vertical text detection model to detect whether the text image contains vertical text;

[0058] Calling a pre-trained long text detection model to detect whether the text image contains long text;

[0059] A pre-trained text language classification model is called to detect the text language of the text image.

[0060] One of the purposes of this embodiment is to provide more accurate input for the subsequent recognition process by identifying the characteristics of the text image, thereby improving the accuracy of text recognition. Among them, multi-dimensional feature detection can refer to performing multi-faceted analysis on the text image before recognizing the text to determine its specific format and language characteristics.

[0061] In the specific implementation of this embodiment, after the text detection model identifies the text area in the image, the vertical text detection model can be first applied to determine whether the text is vertical. If the text is detected to be vertical, the subsequent recognition process will call a model specifically used for vertical text recognition to ensure the accuracy of recognition. Next, the long text detection model is applied to determine whether the text is a long text. If the detection result shows that the text is a long text, a model specially designed for recognizing long text will be used. Finally, the language of the text is determined by the text language classification model so that the recognition model of the corresponding language can be selected for recognition.

[0062] For example, if the text in an image is detected as vertical text, the system can call the vertical text recognition model to process this part of the text. Similarly, if the text is recognized as long text, the system can use the long text recognition model for recognition. In addition, if the text is detected as Chinese, the Chinese recognition model is selected; if it is English, the English recognition model is selected. In this way, the system can not only perform personalized recognition according to the specific situation of the text, but also reduce recognition errors and improve recognition accuracy.

[0063] This implementation method, through early multi-dimensional feature detection, provides more accurate input for the subsequent recognition process, thereby ensuring the reliability of the final recognition result. By selecting the recognition model that best suits the specific format and language characteristics, the system can effectively reduce recognition errors caused by text format or language differences, thereby improving the accuracy and efficiency of the entire machine review system. Therefore, this implementation method can not only improve the recognition accuracy, but also better adapt to the recognition needs of text of different types and formats, ensuring the efficiency and accuracy of image content review.

[0064] Accordingly, based on the above implementation, in some modified implementations, the selecting a corresponding dedicated text recognition model according to the feature detection result to perform text recognition on the text image includes:

[0065] If the feature detection result indicates that the text image is a vertical text image, calling a vertical text recognition model to perform text recognition on the text image;

[0066] If the feature detection result indicates that the text image is a long text image, calling a long text recognition model to perform text recognition on the text image;

[0067] If the feature detection result indicates that the text image is a text image of a specified language, a text recognition model of the specified language is called to perform text recognition on the text image.

[0068] One of the purposes of this embodiment is to improve the accuracy and efficiency of recognition by selecting the most appropriate dedicated text recognition model based on the feature detection results. The dedicated text recognition model may refer to a recognition model optimized for a specific type of text, such as a text model for recognizing vertical text, long text, or a specific language.

[0069] In the specific implementation of this embodiment, after the features of the text image are obtained in the multi-dimensional feature detection stage, the most suitable recognition model can be selected according to the feature detection results. For example, if the detection result shows that the text image is vertical text, the system can call the recognition model that specifically processes vertical text; if the text image is a long text, the recognition model that processes long text is called; if the text image is text in a specified language, the recognition model of the corresponding language is called. In this way, by selecting a model that matches the features of the text image, the accuracy and efficiency of recognition can be improved.

[0070] For example, if the feature detection result shows that the text image is vertical text, the system will call the vertical text recognition model for recognition. The vertical text recognition model usually takes into account the unique layout of vertical text and the connection between characters, so as to more accurately recognize the text content. Similarly, if the text image is a long text, the long text text recognition model is called. This model can better handle the connection and sentence segmentation issues between characters when processing long texts. For text images in different languages, the text recognition models of the corresponding languages ​​are called. These models can better handle the grammatical structure and writing habits of a specific language when recognizing text in that language, thereby improving recognition accuracy.

[0071] In this embodiment, by selecting a dedicated text recognition model based on the feature detection results, the machine review system can provide more accurate recognition results for different types of text images. This can not only reduce recognition errors and improve recognition accuracy and efficiency, but also ensure that different types of text images can receive optimal processing. Therefore, this embodiment can significantly improve the performance of the entire machine review system and ensure the accuracy and efficiency of image content review.

[0072] In some modified implementation manners, before performing multi-dimensional feature detection on the text image, the method further includes:

[0073] A pre-trained text rotation detection model is called to perform rotation detection on the text image, and the text image detected to be rotated is corrected.

[0074] One of the purposes of this embodiment is to detect the rotation of the text image in advance and correct it to ensure that subsequent feature detection and text recognition can be performed on the basis of the correct image.

[0075] The text rotation detection model refers to a model specifically used to detect whether text in a text image is rotated.

[0076] In the specific implementation of this embodiment, before performing multi-dimensional feature detection, the text rotation detection model can be called to perform rotation detection on the text image. If rotation is detected in the text image, the system can correct the image to restore the text to a normal horizontal or vertical state. The correction process usually involves image rotation and / or cropping to ensure that the text is in a standard position for subsequent processing.

[0077] For example, when processing a photo containing tilted text, the text rotation detection model can detect that the text has been rotated and calculate the angle of rotation. The system can then correct the image based on the rotation angle to restore the text to a normal horizontal or vertical state. In this way, subsequent feature detection and text recognition can be performed on a corrected standard image, improving recognition accuracy and efficiency.

[0078] This embodiment adds text rotation detection and correction steps before multi-dimensional feature detection, thereby ensuring that all subsequent processing steps are performed on a standardized image basis. This can not only avoid recognition errors caused by text rotation, but also improve the accuracy and reliability of the entire image review process. Therefore, this embodiment can significantly improve the performance of the machine review system and ensure the accuracy and efficiency of content review.

[0079] In some modified implementations, before performing violation judgment based on all the text recognition results corresponding to the image to be reviewed, the method further includes:

[0080] A pre-trained text error correction model is called to perform error correction processing on all the text recognition results corresponding to the image to be reviewed.

[0081] One of the purposes of this embodiment is to correct potential errors in recognition results through a pre-trained text error correction model, thereby improving the accuracy of text recognition.

[0082] The text error correction model may refer to a model specifically used to identify and correct errors in a text string. The model is usually trained based on machine learning technology and can identify and correct errors generated during the OCR recognition process, such as misrecognition of characters, spelling errors, etc.

[0083] In the specific implementation of this embodiment, after the text in the image is recognized, the system will form all the text recognition results into a text string and send it into a pre-trained text error correction model. This model will analyze the text string, identify possible errors therein, and attempt to correct them. For example, if OCR recognition misidentifies "example" as "exaple", the text error correction model can detect this error and correct it to the correct "example".

[0084] For example, assume that an image contains a text "We are committed to providing high-quality content", but due to some reasons, the OCR recognition result wrongly shows "We are commited to providing high-quality content". At this time, the text error correction model can recognize that "ted" is not a reasonable word in this context and correct it to the correct "mitted" according to the context. In this way, the system can ensure that the final text string is accurate and error-free.

[0085] In this embodiment, by adding a text error correction processing step before the violation judgment, the system can improve the accuracy of the text recognition result and reduce the misjudgment caused by recognition errors. Therefore, this embodiment can ensure the reliability of the subsequent violation judgment and improve the accuracy and efficiency of the entire machine review system.

[0086] In some modified embodiments, the performing the violation judgment according to all the text recognition results corresponding to the image to be reviewed includes:

[0087] Inputting all the text recognition results corresponding to the image to be reviewed into a pre-trained violation text recognition model and using the violation text recognition model to perform the violation judgment; or,

[0088] Performing a violation keyword matching on all the text recognition results corresponding to the image to be reviewed and performing the violation judgment according to the keyword matching result.

[0089] One of the purposes of this embodiment is to ensure that whether there is violation information can be accurately identified by performing the violation judgment on all the recognized text content.

[0090] Among them, the violation text recognition model can refer to a model trained by machine learning technology, which can evaluate whether the text content contains violation information. In the specific implementation of this embodiment, after all the text recognition results are obtained, these results can be passed as input to the violation text recognition model. This model will analyze the text content and evaluate whether these texts contain violation information according to the preset violation criteria. If the model evaluation result indicates that the text has a violation risk, it is determined that the text is in violation.

[0091] In addition, in another embodiment, violation judgment can be made by matching illegal keywords, such as comparing all recognized text strings with a list containing known illegal words. If any text string is found to contain illegal words in the list, the text string is considered to contain illegal information.

[0092] For example, suppose the text recognition result in an image is "This video contains adult content". If the illegal text recognition model is used for judgment, the model will evaluate the content of this text and determine whether it is illegal based on the preset violation standards. If the model evaluation result shows that this text has a risk of violation, then the text is judged to be illegal. For another example, if the judgment is made by illegal keyword matching, the system will compare the recognized text string with a list containing known illegal words. If it is found that the text string contains the illegal word "adult" in the list, then the text string is judged to contain illegal information.

[0093] In this implementation, by using the illegal text recognition model or illegal keyword matching to make illegal judgments, the system can effectively evaluate whether the text content contains illegal information. Therefore, this implementation can improve the accuracy and efficiency of image review and ensure the security and compliance of platform content.

[0094] In some modified implementations, before performing violation judgment based on all the text recognition results corresponding to the image to be reviewed, the method further includes:

[0095] Calling a pre-trained computer vision recognition model to identify scene element information in the image to be reviewed;

[0096] The making violation judgment according to all the text recognition results corresponding to the image to be reviewed includes:

[0097] A violation judgment is performed based on the overall information consisting of all the text recognition results and the scene element information corresponding to the image to be reviewed to obtain a violation judgment result.

[0098] One of the purposes of this embodiment is to comprehensively evaluate whether an image contains illegal content by combining the text content and non-text scene element information in the image, thereby improving the accuracy and comprehensiveness of violation judgment.

[0099] Among them, the computer vision recognition (CVR) model can refer to an algorithm model trained by machine learning, which is used to identify various scene elements in an image, including people, animals, environments, objects, actions, colors, etc., and convert them into understandable and processable information. The CVR model is usually based on deep learning technology, such as convolutional neural networks (CNNs), which can accurately identify various scene elements in an image. In this embodiment, after text recognition, the computer vision recognition model can be called to analyze the scene elements in the image, and then the recognized text content is combined with the scene element information to form a complete image description, and then the overall information is used to make a violation judgment to determine whether the image contains illegal content.

[0100] For example, suppose an image contains a text "Illegal items for sale" and there is a scene in the image that displays prohibited items. First, the text "Illegal items for sale" in the image is identified through OCR technology. Then, a computer vision recognition model is used to analyze the scene elements in the image and identify the prohibited items displayed in the image. Then, the text content "Illegal items for sale" is combined with the scene element information "Displaying prohibited items" to make a comprehensive violation judgment. In this way, the system can not only identify the illegal content in the text, but also further confirm whether the image contains illegal information by analyzing the scene in the image.

[0101] In this embodiment, by combining text content and scene element information to make violation judgments, the system can more comprehensively evaluate the overall content of the image, thereby improving the accuracy and reliability of violation judgments. Therefore, this embodiment can significantly improve the performance of the machine review system and ensure the accuracy and efficiency of content review.

[0102] To facilitate understanding, the above-mentioned image review method of the present application is further explained with reference to a specific example.

[0103] In a specific example, after the image is input, text detection is first performed. After the detection is completed, a rotation model will be applied to determine whether the text is rotated, and whether it is vertical or long text. After these judgments are completed, the text will be classified into different languages, because the classified text will be fed into the corresponding language model to improve the recognition accuracy. For example, long text will be recognized using a special long text model, and several rows of text will be recognized using a special several rows of text model. The results recognized by these models will be fed into the text error correction model to identify and correct possible algorithm recognition errors. After confirmation, the OCR output is finally generated.

[0104] However, even after the OCR output is completed, it is not entirely certain whether the text content violates the rules. Therefore, it is necessary to further access the text violation detection link, including the violation keyword matching model and the violation text recognition model. The violation keyword matching model will determine whether the text string contains illegal words, and for illegal words, it will be directly submitted for review. The illegal text recognition model evaluates whether the text content is illegal or vulgar. If the evaluation result exceeds a specific threshold, it will also be submitted for review. In order to pursue higher recognition accuracy, this solution adds auxiliary models and correction models on the basis of the existing OCR model to improve the accuracy of the recognition results.

[0105] For details, please refer to Figure 2 , which shows a flowchart of a specific example of an image review method provided by an embodiment of the present application. The image review method shown in the figure may include the following steps:

[0106] S1. Obtain the image to be reviewed;

[0107] S2, calling a text detection model to perform text detection on the image to be reviewed, and cropping the image to be reviewed based on the detected text to obtain a text image;

[0108] S3, calling a text rotation detection model to perform rotation detection and rotation correction on the text image;

[0109] S4, calling the vertical text detection model to detect whether the text image contains vertical text, if so, jump to step S702, if not, jump to S5;

[0110] S5, calling the long text detection model to detect whether the text image contains long text, if so, jump to step S703, if not, jump to S6;

[0111] S6, calling a text language classification model to classify the text image to obtain a classification result;

[0112] S7, calling a text recognition model to perform text recognition on the text image to obtain a text recognition result; wherein different solutions are adopted for different situations:

[0113] S701: For the text image outputted in S6, a text recognition model of a corresponding language category is selected for recognition according to the classification result. The text recognition model includes but is not limited to a Chinese recognition model, an English recognition model, and other language recognition models.

[0114] S702: for the vertical text image detected in S4, calling a vertical text recognition model to perform recognition;

[0115] S703: for the long text image detected in S5, calling the long text recognition model to perform recognition;

[0116] S8, calling a text error correction model to perform text error correction on the text recognition result output by S7;

[0117] S9, merging all text recognition results corresponding to the image to be reviewed into a text string, and then outputting the text string corresponding to the image to be reviewed, wherein the synthesis of the text string and then performing violation judgment can provide more comprehensive context information, which helps to improve the accuracy of violation judgment;

[0118] S10, performing violation judgment on the text string and outputting the violation judgment result, including two methods:

[0119] Method 1: calling the illegal keyword matching model on the text string to match the illegal keywords. If the illegal keywords are matched, it is determined that the text string contains illegal information, and the illegal judgment information is output;

[0120] Method 2: calling a violation text recognition model to conduct violation review on the text string, and the violation text recognition model outputs violation risk level information corresponding to the text string;

[0121] S11. Process the image to be reviewed according to the violation judgment result, including releasing, deleting, hiding, transferring to manual review, etc. If the image to be reviewed is judged to be in violation, the image to be reviewed can be directly deleted, hidden, or transferred to manual review according to the preset violation image processing rules. The purpose of transferring to manual review is to further confirm whether it is in violation through manual review to avoid misjudgment.

[0122] It should be noted that in the above embodiments of the present application, a large number of models are introduced, including text detection models, text rotation detection models, vertical text detection models, long text detection models, text language classification models, special text recognition models, vertical text recognition models, long text recognition models, text error correction models, illegal text recognition models and computer vision recognition models. In specific implementation, the above models can be applied to the present application scheme directly or after fine-tuning and training using existing models provided by the prior art. The embodiments of the present application do not limit the specific content of the text detection model.

[0123] Based on the same inventive concept, an image review device is also provided in an embodiment of the present application. The method corresponding to the device may be the image review method in the aforementioned embodiment, and its principle of solving the problem is similar to that of the method. The image review device provided in an embodiment of the present application can implement the aforementioned image review method, and the image review device can be implemented by software, hardware, or a combination of software and hardware. For example, the image review device may include integrated or separate functional modules or units to perform the corresponding steps in the aforementioned methods. Since the device embodiment is basically similar to the method embodiment, the description is relatively simple, and the relevant parts can be referred to the partial description of the method embodiment. The device embodiment described below is merely illustrative. Please refer to Figure 3 , the image review device may include:

[0124] The text detection module 101 is used to call the text detection model to perform text detection on the image to be reviewed, and to crop the image to be reviewed according to the detected text to obtain at least one text image;

[0125] The text recognition module 102 is used to perform multi-dimensional feature detection on each of the text images, and select a corresponding dedicated text recognition model to perform text recognition on the text image according to the feature detection result to obtain a text recognition result;

[0126] A violation judgment module 103 is used to perform violation judgment according to all the text recognition results corresponding to the image to be reviewed to obtain a violation judgment result;

[0127] The illegal image processing module 104 is used to process the to-be-reviewed image that is judged to be illegal according to a preset illegal image processing rule.

[0128] In some modified implementations, the text recognition module 102 includes at least one of the following feature detection units:

[0129] A vertical feature detection unit, used to call a pre-trained vertical text detection model to detect whether the text image contains vertical text;

[0130] A long text feature detection unit, used to call a pre-trained long text detection model to detect whether the text image contains a long text;

[0131] The language feature detection unit is used to call a pre-trained text language classification model to detect the text language of the text image.

[0132] In some modified implementations, the text recognition module 102 includes at least one of the following text recognition units:

[0133] A vertical text recognition unit, configured to call a vertical text recognition model to perform text recognition on the text image if the feature detection result indicates that the text image is a vertical text image;

[0134] A long text recognition unit, configured to call a long text recognition model to perform text recognition on the text image if the feature detection result indicates that the text image is a long text image;

[0135] The designated language text recognition unit is used to call the designated language text recognition model to perform text recognition on the text image if the feature detection result indicates that the text image is a designated language text image.

[0136] In some modified implementations, the text recognition module 102 further includes:

[0137] The rotation correction unit is used to call a pre-trained text rotation detection model to perform rotation detection on the text image, and correct the text image that is detected to be rotated.

[0138] In some variations, the device further comprises:

[0139] The text error correction model is used to call a pre-trained text error correction model to perform error correction processing on all the text recognition results corresponding to the image to be reviewed.

[0140] In some modified implementations, the violation determination module 103 includes:

[0141] A model judgment unit is used to input all the text recognition results corresponding to the image to be reviewed into a pre-trained illegal text recognition model, and use the illegal text recognition model to perform illegal judgment; or

[0142] The keyword judgment unit is used to match the illegal keywords of all the text recognition results corresponding to the image to be reviewed, and make illegal judgments based on the keyword matching results.

[0143] In some variations, the device further comprises:

[0144] A scene element recognition module, used to call a pre-trained computer vision recognition model to recognize scene element information in the image to be reviewed;

[0145] The violation judgment module 103 includes:

[0146] The overall judgment unit is used to make a violation judgment based on the overall information composed of all the text recognition results and the scene element information corresponding to the image to be reviewed, and obtain a violation judgment result.

[0147] The image review device provided in the embodiment of the present application is based on the same inventive concept as the image review method provided in the aforementioned embodiment of the present application and has the same beneficial effects.

[0148] Based on the same inventive concept, an electronic device is also provided in an embodiment of the present application, and the method corresponding to the electronic device may be the image review method in the aforementioned embodiment, and its principle of solving the problem is similar to that of the method. The electronic device provided in an embodiment of the present application includes: at least one processor; and a memory connected to the at least one processor in communication; wherein the memory stores instructions executable by the at least one processor, and the instructions are executed by the at least one processor so that the at least one processor can execute the image review methods and / or technical solutions of the aforementioned multiple embodiments of the present application.

[0149] The electronic device may be a user device, or a device formed by integrating a user device and a network device through a network, or may be an application running on the above device. The user device includes but is not limited to various terminal devices such as computers, mobile phones, tablet computers, smart watches, and bracelets. The network device includes but is not limited to network hosts, single network servers, multiple network server sets, or cloud computing-based computer sets, which can be used to implement some processing functions when setting an alarm. Here, the cloud is composed of a large number of hosts or network servers based on cloud computing, where cloud computing is a type of distributed computing, a virtual computer composed of a group of loosely coupled computer sets.

[0150] Figure 4 The structure of an electronic device suitable for implementing the method and / or technical solution in the embodiment of the present application is shown, and the electronic device 1200 includes a central processing unit (CPU, Central Processing Unit) 1201, which can perform various appropriate actions and processes according to the program stored in the read-only memory (ROM, Read Only Memory) 1202 or the program loaded from the storage part 1208 to the random access memory (RAM, Random Access Memory) 1203. In RAM 1203, various programs and data required for system operation are also stored. CPU 1201, ROM 1202 and RAM1203 are connected to each other through bus 1204. Input / output (I / O, Input / Output) interface 1205 is also connected to bus 1204.

[0151] The following components are connected to the I / O interface 1205: an input section 1206 including a keyboard, a mouse, a touch screen, a microphone, an infrared sensor, etc.; an output section 1207 including a cathode ray tube (CRT), a liquid crystal display (LCD), an LED display, an OLED display, etc., and a speaker, etc.; a storage section 1208 including one or more computer-readable media such as a hard disk, an optical disk, a magnetic disk, a semiconductor memory, etc.; and a communication section 1209 including a network interface card such as a LAN (Local Area Network) card, a modem, etc. The communication section 1209 performs communication processing via a network such as the Internet.

[0152] In particular, the methods and / or embodiments in the embodiments of the present application may be implemented as computer software programs. For example, the embodiments disclosed in the present application include a computer program product, which includes a computer program carried on a computer-readable medium, and the computer program includes program code for executing the method shown in the flowchart. When the computer program is executed by the central processing unit (CPU) 1201, the above functions defined in the method of the present application are executed.

[0153] Another embodiment of the present application further provides a computer-readable storage medium having computer program instructions stored thereon, wherein the computer program instructions can be executed by a processor to implement the methods and / or technical solutions of any one or more embodiments of the present application described above.

[0154] Specifically, the present embodiment may adopt any combination of one or more computer-readable media. The computer-readable medium may be a computer-readable signal medium or a computer-readable storage medium. The computer-readable storage medium may be, for example, but not limited to, an electrical, magnetic, optical, electromagnetic, infrared, or semiconductor system, device or device, or any combination thereof. More specific examples (non-exhaustive list) of computer-readable storage media include: an electrical connection with one or more wires, a portable computer disk, a hard disk, a random access memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM or flash memory), an optical fiber, a portable compact disk read-only memory (CD-ROM), an optical storage device, a magnetic storage device, or any suitable combination thereof. In the present application, a computer-readable storage medium may be any tangible medium containing or storing a program, which may be used by an instruction execution system, device or device or used in combination therewith.

[0155] Computer readable signal media may include a data signal propagated in baseband or as part of a carrier wave, which carries a computer readable program code. Such propagated data signals may take a variety of forms, including, but not limited to, electromagnetic signals, optical signals, or any suitable combination of the above. Computer readable signal media may also be any computer readable medium other than a computer readable storage medium, which may send, propagate, or transmit a program for use by or in conjunction with an instruction execution system, apparatus, or device.

[0156] Program code embodied on a computer readable medium may be transmitted using any appropriate medium, including but not limited to wireless, wireline, optical fiber cable, RF, etc., or any suitable combination of the foregoing.

[0157] Computer program code for performing the operations of the present application may be written in one or more programming languages ​​or a combination thereof, including object-oriented programming languages, such as Java, Smalltalk, C++, and conventional procedural programming languages, such as "C" or similar programming languages. The program code may be executed entirely on the user's computer, partially on the user's computer, as a separate software package, partially on the user's computer and partially on a remote computer, or entirely on a remote computer or server. In the case of a remote computer, the remote computer may be connected to the user's computer through any type of network, including a local area network (LAN) or a wide area network (WAN), or may be connected to an external computer (e.g., via the Internet using an Internet service provider).

[0158] The flow chart or block diagram in the accompanying drawings shows the possible architecture, function and operation of the equipment, method and computer program product according to various embodiments of the present application. In this regard, each square box in the flow chart or block diagram can represent a module, a program segment or a part of a code, and the module, the program segment or a part of the code contains one or more executable instructions for realizing the specified logical function. It should also be noted that in some implementations as replacements, the functions marked in the square box can also occur in a sequence different from that marked in the accompanying drawings. For example, two square boxes represented in succession can actually be executed substantially in parallel, and they can sometimes be executed in the opposite order, depending on the functions involved. It should also be noted that each square box in the block diagram and / or flow chart, and the combination of the square boxes in the block diagram and / or flow chart can be implemented with a dedicated system for hardware that performs a specified function or operation, or can be implemented with a combination of dedicated hardware and computer instructions.

[0159] Those skilled in the art can clearly understand that, for the convenience and brevity of description, the specific working processes of the systems, devices and units described above can refer to the corresponding processes in the aforementioned method embodiments and will not be repeated here.

[0160] In the several embodiments provided in the present application, it should be understood that the disclosed systems, devices and methods can be implemented in other ways. For example, the device embodiments described above are only schematic. For example, the division of the units is only a logical function division. There may be other division methods in actual implementation. For example, multiple units or page components can be combined or integrated into another system, or some features can be ignored or not executed. Another point is that the mutual coupling or direct coupling or communication connection shown or discussed can be an indirect coupling or communication connection through some interfaces, devices or units, which can be electrical, mechanical or other forms.

[0161] The units described as separate components may or may not be physically separated, and the components shown as units may or may not be physical units, that is, they may be located in one place or distributed on multiple network units. Some or all of the units may be selected according to actual needs to achieve the purpose of the solution of this embodiment.

[0162] In addition, each functional unit in each embodiment of the present application may be integrated into one processing unit, or each unit may exist physically separately, or two or more units may be integrated into one unit. The above-mentioned integrated unit may be implemented in the form of hardware or in the form of hardware plus software functional units.

[0163] The above-mentioned integrated unit implemented in the form of a software functional unit can be stored in a computer-readable storage medium. The above-mentioned software functional unit is stored in a storage medium, including a number of instructions for a computer device (which can be a personal computer, a server, or a network device, etc.) or a processor to perform some steps of the method described in each embodiment of the present application. The aforementioned storage medium includes: U disk, mobile hard disk, read-only memory (Read-Only Memory, ROM), random access memory (Random Access Memory, RAM), disk or optical disk and other media that can store program codes.

[0164] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present application, rather than to limit it. Although the present application has been described in detail with reference to the aforementioned embodiments, those skilled in the art should understand that they can still modify the technical solutions described in the aforementioned embodiments, or make equivalent replacements for some of the technical features therein. However, these modifications or replacements do not deviate the essence of the corresponding technical solutions from the spirit and scope of the technical solutions of the embodiments of the present application.

[0165] In addition, it is clear that the word "comprising" does not exclude other units or steps, and the singular does not exclude the plural. Multiple units or devices stated in a device claim can also be implemented by one unit or device through software or hardware. The words first, second, etc. are used to indicate names, and do not indicate any particular order.

Claims

1. An image review method, wherein: The method comprises: Calling a text detection model to perform text detection on the image to be reviewed, and cropping the image to be reviewed according to the detected text to obtain at least one text image; For each of the text images, multi-dimensional feature detection is performed on the text image, and a corresponding dedicated text recognition model is selected according to the feature detection result to perform text recognition on the text image to obtain a text recognition result; Performing violation judgment according to all the text recognition results corresponding to the image to be reviewed to obtain a violation judgment result; The to-be-audited image judged to be in violation is processed according to preset violation image processing rules.

2. The image review method according to claim 1, wherein: The performing multi-dimensional feature detection on the text image includes at least one of the following: Calling a pre-trained vertical text detection model to detect whether the text image contains vertical text; Calling a pre-trained long text detection model to detect whether the text image contains long text; A pre-trained text language classification model is called to detect the text language of the text image.

3. The image review method according to claim 2, wherein: The step of selecting a corresponding dedicated text recognition model according to the feature detection result to perform text recognition on the text image includes: If the feature detection result indicates that the text image is a vertical text image, calling a vertical text recognition model to perform text recognition on the text image; If the feature detection result indicates that the text image is a long text image, calling a long text recognition model to perform text recognition on the text image; If the feature detection result indicates that the text image is a text image of a specified language, a text recognition model of the specified language is called to perform text recognition on the text image.

4. The image review method according to claim 1, wherein: Before performing multi-dimensional feature detection on the text image, the method further includes: A pre-trained text rotation detection model is called to perform rotation detection on the text image, and the text image detected to be rotated is corrected.

5. The image review method according to claim 1, wherein: Before making violation judgment according to all the text recognition results corresponding to the image to be reviewed, the method further includes: A pre-trained text error correction model is called to perform error correction processing on all the text recognition results corresponding to the image to be reviewed.

6. The image review method according to claim 1, wherein: The making violation judgment according to all the text recognition results corresponding to the image to be reviewed includes: Input all the text recognition results corresponding to the image to be reviewed into a pre-trained illegal text recognition model, and use the illegal text recognition model to make a violation judgment; or All the text recognition results corresponding to the image to be reviewed are matched with illegal keywords, and illegal judgment is made based on the keyword matching results.

7. The image review method according to claim 1, wherein: Before making violation judgment according to all the text recognition results corresponding to the image to be reviewed, the method further includes: Calling a pre-trained computer vision recognition model to identify scene element information in the image to be reviewed; The determining of violation according to all the text recognition results corresponding to the image to be reviewed includes: A violation judgment is performed based on the overall information consisting of all the text recognition results and the scene element information corresponding to the image to be reviewed to obtain a violation judgment result.

8. An image review device, wherein: The device comprises: A text detection module, used to call a text detection model to perform text detection on the image to be reviewed, and to crop the image to be reviewed according to the detected text to obtain at least one text image; A text recognition module is used to perform multi-dimensional feature detection on each of the text images, and select a corresponding dedicated text recognition model to perform text recognition on the text image according to the feature detection result to obtain a text recognition result; A violation judgment module, used to make a violation judgment based on all the text recognition results corresponding to the image to be reviewed, and obtain a violation judgment result; The illegal image processing module is used to process the to-be-audited image that is judged to be illegal according to a preset illegal image processing rule.

9. An electronic device, comprising: at least one processor; as well as a memory communicatively connected to the at least one processor; wherein, The memory stores instructions that can be executed by the at least one processor, and the instructions are executed by the at least one processor to enable the at least one processor to perform the method according to any one of claims 1 to 7.

10. A computer readable medium having computer program instructions stored thereon, wherein the computer program instructions can be executed by a processor to implement the method according to any one of claims 1 to 7.

Images