Illegal content detection system for display equipment terminal and use method of illegal content detection system

Abstract

The invention relates to a violation content detection system for a display equipment terminal and a use method thereof, which are applied to the field of information processing, and the system comprises a violation multi-mode extraction module, a multi-mode violation content detection algorithm module, a violation content judgment module, display screen external connection management and control equipment and a display terminal for playing a video stream, after a display terminal receives a video stream, data in various modes can be processed and analyzed, including texts, audios, videos, high-dimensional semantics and paragraph information, so that quick and accurate detection of violation contents is realized, playing contents are automatically audited without manual intervention, and the playing efficiency is improved. The method comprises the following steps: detecting risk content of a display terminal, triggering preset risk management and control measures when the risk content is detected, and carrying out state monitoring on each USB port and HDMI port which are set to be in an access occupation state in the display terminal in an initial state so as to judge whether the USB port and the HDMI port are pulled out or not, so that safe use of equipment is ensured, and malicious access is prevented.

Description

Technical Field

[0001] The present invention relates to an illegal content detection system for a display device terminal, and in particular to an illegal content detection system for a display device terminal applied in the field of information processing and a method of using the system. Background Art

[0002] At present, most of the display devices such as outdoor display screens, LED screens, TV screens, and splicing screens on the market do not conduct illegal content inspections or simple inspections, and there are security risks in the release of near-source content on display device terminals. The current technology used to detect illegal content is mainly through single-mode content detection such as text, pictures, and audio. For example, for video content, one solution can use voice recognition technology to match sensitive words and keywords to analyze whether the audio content in the video is illegal; for pictures, machine learning models can be used to identify illegal content such as pornography, violence, and malicious advertising.

[0003] The specification of Chinese invention patent CN111921204B discloses a method, apparatus, device and storage medium for controlling a cloud application. During a user's trial of a cloud application, the user's trial is terminated proactively when an illegal operation is detected, thereby preventing the user's illegal operation from causing the system account of the cloud device to be unable to be used normally, reducing the burden on the cloud device and operation and maintenance personnel, and controlling the cloud application by monitoring the display content of the application running screen, without making major changes to the cloud application itself, thereby improving the efficiency of controlling the cloud application.

[0004] The existing implementation of illegal content detection mainly relies on manual review of content detection on the publishing end, supplemented by keyword matching program detection. The illegal content detection method is relatively simple, with limited capabilities and a high missed detection rate. In addition, the content on the playback end is not detected and controlled. Even if illegal content is detected, there is a lack of effective cutting methods to prevent the spread of illegal content. Summary of the invention

[0005] In view of the above-mentioned prior art, the technical problem to be solved by the present invention is to perform efficient, comprehensive and rapid detection of illegal content on the display terminal, and to perform effective transmission cut-off processing to prevent malicious insertion during playback.

[0006] To solve the above problems, the present invention provides an illegal content detection system for a display device terminal, which is used to detect illegal content in a received video stream, and includes an illegal multimodal extraction module, a multimodal illegal content detection algorithm module, an illegal content judgment module, a display screen external connection management and control device, and a display terminal for playing a video stream;

[0007] The violation multimodal extraction module is used to extract the features of speech, text and image modalities in the video stream;

[0008] The multimodal illegal content detection algorithm module will receive the features extracted by the illegal multimodal extraction module and process them in parallel;

[0009] The illegal content judgment module receives the result of the multimodal illegal content detection algorithm module and makes a judgment. If it is judged to be illegal content, it will send a power-off command to the display screen external control device;

[0010] The display screen external control device judges the instructions of the module based on the received illegal content and performs real-time detection on whether each USB and HDMI port on the surface of the display terminal is unplugged, controls the relay in the display terminal, turns off the power switch of the display terminal, and implements power-off operation.

[0011] In the above-mentioned illegal content detection system for display device terminals, it is possible to detect in real time whether the multimodal content and high-dimensional text semantics and voice paralanguage information of the output device terminal contain illegal content, and monitor the status of each USB and HDMI port of the display terminal to prevent malicious insertion.

[0012] As a further improvement of the present application, the illegal content judgment module includes a text illegal content judgment unit, an audio illegal content judgment unit and a video illegal content judgment unit, which comprehensively detects whether there is illegal information in the features extracted by the illegal multimodal extraction module from the three modalities of text, picture and audio;

[0013] The text violation content judgment unit analyzes and detects whether there is any violation in the text features;

[0014] The audio illegal content judgment unit can not only identify and judge whether there is illegal content in the voice features, but also judge whether there is illegal content in the paralinguistic information of the voice features;

[0015] The video illegal content judgment unit judges whether there is illegal content in the image features.

[0016] As a further improvement of the present application, the display terminal includes a display screen and a relay, and the power-on state of the relay controls the opening and closing of the display terminal.

[0017] As a further improvement of the present application, the display terminal also includes multiple USB and HDMI ports arranged on the surface of the display terminal, and the interior of each USB and HDMI port is connected to the signal of the external control device of the display screen by installing a hardware sensor. In the initial state, each USB and HDMI port is displayed as an access placeholder state.

[0018] As a further improvement of the present application, the audio illegal content judgment unit includes a speech recognition algorithm block and a speech classification algorithm block, wherein the speech recognition algorithm block translates the sound into text and matches the keyword detection to judge the illegal content, and the speech classification algorithm block performs paralinguistic information content detection to judge whether the sound involves violence, pornography and terrorist content.

[0019] As a further improvement of the present application, the text illegal content judgment unit includes a text recognition block and a text analysis block, wherein the text recognition module is used to detect whether there is illegal content in the text features, and the text analysis block is used to analyze the text semantic content and determine whether there is illegal content in the text semantics.

[0020] A method for using a system for detecting illegal content for a display device terminal includes the following steps:

[0021] S1: The video stream is transmitted to the illegal multi-modal extraction module, and the features of the received video stream in three modes, namely text, picture and audio, are extracted;

[0022] S2, transmitting the features extracted in S1 to the multimodal illegal content detection algorithm module, and transmitting the extracted text features, image features, and audio features to the text illegal content judgment unit, the audio illegal content judgment unit, and the video illegal content judgment unit, respectively, to judge whether there is illegal content;

[0023] S3. Before the display terminal plays the video, each USB and HDMI port on the display terminal is controlled by the display screen external control device to be in the access occupied state;

[0024] S4. When S2 detects that there is no illegal content in the video stream and each USB and HDMI port in S3 is not unplugged, the display terminal plays the video stream normally. Otherwise, the control relay of the display screen external control device is powered off and the display terminal stops playing.

[0025] S2 also includes the following steps:

[0026] S21, the text violation content judgment unit first uses the text recognition block to identify whether there is any violation content in the received text features. If not, the text analysis block performs text semantic analysis on the text features to determine whether there is any violation in the text semantics;

[0027] S22, the audio violation content judgment unit transmits the received audio features to the speech recognition algorithm block and the speech classification algorithm block at the same time. The speech recognition algorithm block translates the audio features into text and matches the keyword detection. The recognized text will be semantically analyzed to see if it contains sensitive violation information. The speech recognition algorithm block uses the sound classification algorithm to detect the audio features to determine whether there is violence, pornography, and terror. If there is, it indicates that there is a violation.

[0028] S23. The video illegal content judgment unit will extract image features from frames, and judge whether the picture contains illegal content based on the image content. At the same time, it will use OCR technology to recognize the text in the picture. After matching keyword detection, the recognized text will extract text language features and then analyze it to see whether it contains sensitive information.

[0029] In summary, after receiving the video stream, the display terminal can process and analyze data in multiple modes, including text, audio and video, as well as high-dimensional semantic and paralinguistic information, to achieve fast and accurate detection of illegal content, automatically review the playback content without human intervention, and trigger preset risk control measures when risky content is detected. In the initial state, each USB and HDMI port in the display terminal that is set to be connected in a placeholder state is monitored to determine whether the USB and HDMI ports are unplugged, thereby ensuring the safe use of the device and preventing malicious access. BRIEF DESCRIPTION OF THE DRAWINGS

[0030] Figure 1 This is a system composition diagram of this application;

[0031] Figure 2 This is the system flow chart of this application. DETAILED DESCRIPTION

[0032] Two implementation modes of the present application are described in detail below with reference to the accompanying drawings.

[0033] The first implementation method:

[0034] Figure 1 A system for detecting illegal contents for a display device terminal is shown, which is used to detect illegal contents in a received video stream, and includes an illegal multimodal extraction module, a multimodal illegal content detection algorithm module, an illegal content judgment module, a display screen external connection control device, and a display terminal for playing a video stream;

[0035] The violation multimodal extraction module is used to extract the features of speech, text and image modalities in the video stream;

[0036] The multimodal illegal content detection algorithm module will receive the features extracted by the illegal multimodal extraction module and process them in parallel;

[0037] The illegal content judgment module receives the result of the multimodal illegal content detection algorithm module and makes a judgment. If it is judged to be illegal content, it will send a power-off command to the display screen external control device;

[0038] The display screen external control device judges the instructions of the module based on the received illegal content and performs real-time detection on whether each USB and HDMI port on the surface of the display terminal is unplugged, controls the relay in the display terminal, turns off the power switch of the display terminal, and implements power-off operation.

[0039] Specifically, according to actual needs, parameters of the multimodal illegal content detection algorithm module are set at the video playback end, such as setting sensitive text keywords for detection, etc., and it can process and analyze data in multiple modes, including text, audio and video, as well as high-dimensional semantics (that is, the analysis and understanding of text content) and paralinguistic information (that is, other audio content that can be converted and recognized into text, such as some specific audio features, such as abnormal sounds such as screaming and explosions), so as to achieve fast and accurate detection of illegal content. After detecting illegal content, the display screen external control device is used in time to block video playback, and the USB and HDMI ports of the display device terminal can be supervised to avoid malicious playback.

[0040] The illegal content judgment module includes a text illegal content judgment unit, an audio illegal content judgment unit and a video illegal content judgment unit, which comprehensively detects whether there is illegal information in the features extracted by the illegal multimodal extraction module from the three modalities of text, picture and audio;

[0041] The text violation content judgment unit analyzes and detects whether there is any violation in the text features;

[0042] The audio illegal content judgment unit can not only identify and judge whether there is illegal content in the voice features, but also judge whether there is illegal content in the paralinguistic information of the voice features;

[0043] The video illegal content judgment unit judges whether there is illegal content in the image features.

[0044] The display terminal includes a display screen and a relay, and the power-on state of the relay controls the opening and closing of the display terminal.

[0045] The audio illegal content judgment unit includes a speech recognition algorithm block and a speech classification algorithm block. The speech recognition algorithm block translates the sound into text and then matches the keyword detection to judge the illegal content. The speech classification algorithm block performs paralinguistic information content detection to judge whether the sound involves violence, pornography and terrorist content.

[0046] The text violation content judgment unit includes a text recognition block and a text analysis block, wherein the text recognition module is used to detect whether there is any violation content in the text features, and the text analysis block is used to analyze the text semantic content and judge whether there is any violation content in the text semantics.

[0047] Specifically, multimodal illegal content detection is more comprehensive than single-modal illegal content detection. It can detect in real time whether the multimodal content of the output device terminal and the high-dimensional text semantics and speech paralanguage information contain illegal content, thereby improving the accuracy and reducing the missed detection rate.

[0048] Figure 2 A method for using a system for detecting illegal content for a display device terminal is shown, comprising the following steps:

[0049] S1: The video stream is transmitted to the illegal multi-modal extraction module, and the features of the received video stream in three modes, namely text, picture and audio, are extracted;

[0050] S2, transmitting the features extracted in S1 to the multimodal illegal content detection algorithm module, and transmitting the extracted text features, image features, and audio features to the text illegal content judgment unit, the audio illegal content judgment unit, and the video illegal content judgment unit, respectively, to judge whether there is illegal content;

[0051] S3. Before the display terminal plays the video, each USB and HDMI port on the display terminal is controlled by the display screen external control device to be in the access occupied state;

[0052] S4. When S2 detects that there is no illegal content in the video stream and each USB and HDMI port in S3 is not unplugged, the display terminal plays the video stream normally. Otherwise, the control relay of the display screen external control device is powered off and the display terminal stops playing.

[0053] S2 also includes the following steps:

[0054] S21, the text violation content judgment unit first uses the text recognition block to identify whether there is any violation content in the received text features. If not, the text analysis block performs text semantic analysis on the text features to determine whether there is any violation in the text semantics;

[0055] S22, the audio violation content judgment unit transmits the received audio features to the speech recognition algorithm block and the speech classification algorithm block at the same time. The speech recognition algorithm block translates the audio features into text and matches the keyword detection. The recognized text will be semantically analyzed to see if it contains sensitive violation information. The speech recognition algorithm block uses the sound classification algorithm to detect the audio features to determine whether there is violence, pornography, and terror. If there is, it indicates that there is a violation.

[0056] S23. The video illegal content judgment unit will extract image features from frames, and judge whether the picture contains illegal content based on the image content. At the same time, it will use OCR technology to recognize the text in the picture. After matching keyword detection, the recognized text will extract text language features and then analyze it to see whether it contains sensitive information.

[0057] Specifically, during the playback process, once the system detects content that may be risky, it will automatically trigger the risk control mechanism (that is, the display screen external control device). The display screen external control device can respond quickly and cut off the power supply of the playback device, thereby effectively preventing the further spread of bad content, improving supervision efficiency, saving labor costs, and ensuring the timeliness and accuracy of supervision. Whether in enterprises, schools, public places or homes, it can play an important role in various playback scenarios.

[0058] The second implementation method:

[0059] The display terminal also includes multiple USB and HDMI ports arranged on the surface of the display terminal, and the interior of each USB and HDMI port is connected to the signal of the external control device of the display screen by installing a hardware sensor. In the initial state, each USB and HDMI port is displayed as an access placeholder state.

[0060] Specifically, the display terminal is suitable for various output device terminals, such as computers, monitors, projectors, etc., with a wide range of applications. In the initial state, each USB and HDMI port of the display terminal is set to an access placeholder state. Therefore, in the subsequent playback process, if someone is maliciously inserted into the playback device to play illegal content, it is necessary to unplug one of the USB and HDMI ports to destroy the access placeholder state. At this time, the control relay of the display screen external control device is powered off to prevent the display terminal from continuing to play the video. By real-time monitoring the status of each USB and HDMI port of the display terminal, malicious access to the video playback device can be effectively prevented.

[0061] In summary, this application performs illegal content detection and control at the video playback end (that is, the display terminal), because in the prior art, the data transmission path, whether it is ordinary network data transmission or encrypted transmission content, may be maliciously tampered with during the transmission process. Although encrypted data transmission improves security to a certain extent, it still cannot completely eliminate the risk of tampering. After decryption at the end of the device, there is still the possibility of tampering, so the detection and control at the playback end has obvious advantages;

[0062] After the display terminal receives the video stream, it can process and analyze data in multiple modes, including text, audio and video, as well as high-dimensional semantic and paralinguistic information, to achieve fast and accurate detection of illegal content. It can automatically review the playback content without human intervention, and trigger preset risk control measures when risky content is detected. It also monitors the status of each USB and HDMI port in the display terminal that is set to be connected and occupied in the initial state to determine whether the USB and HDMI ports are unplugged, ensure the safe use of the device, and prevent malicious access.

[0063] In view of current practical needs, the above-mentioned implementation mode adopted in this application is not limited to the scope of protection. Various changes made within the knowledge scope of technical personnel in this field without departing from the concept of this application still fall within the scope of protection of the present invention.

Claims

1. A system for detecting illegal content for a display device terminal, used for detecting illegal content in a received video stream, characterized in that: It includes a violation multimodal extraction module, a multimodal violation content detection algorithm module, a violation content judgment module, a display screen external control device and a display terminal for playing video streams; The illegal multimodal extraction module is used to extract features of speech, text and picture modalities in the video stream; The multimodal illegal content detection algorithm module receives the features extracted by the illegal multimodal extraction module and processes them in parallel; The illegal content judgment module receives the result of the multimodal illegal content detection algorithm module and makes a judgment. If it is judged to be illegal content, it will issue a power-off instruction to the display screen external control device; The display screen external control device detects in real time whether each USB and HDMI port on the surface of the display terminal is unplugged according to the instructions of the received illegal content judgment module, controls the relay in the display terminal, turns off the power switch of the display terminal, and implements the power-off operation.

2. The illegal content detection system for a display device terminal according to claim 1, characterized in that: The illegal content judgment module includes a text illegal content judgment unit, an audio illegal content judgment unit and a video illegal content judgment unit, which comprehensively detects whether there is illegal information in the features extracted by the illegal multimodal extraction module from the three modalities of text, picture and audio; The text violation content judgment unit analyzes and detects whether there is any violation in the text features; The audio illegal content judgment unit can not only identify and judge whether there is illegal content in the voice features, but also judge whether there is illegal content in the paralinguistic information of the voice features; The video illegal content judgment unit judges whether there is illegal content in the image features.

3. The illegal content detection system for a display device terminal according to claim 1, characterized in that: The display terminal includes a display screen and a relay, and the power-on state of the relay controls the opening and closing of the display terminal.

4. The illegal content detection system for a display device terminal according to claim 3, characterized in that: The display terminal also includes multiple USB and HDMI ports arranged on the surface of the display terminal, and the interior of each USB and HDMI port is connected to the display screen external control device signal by installing a hardware sensor. In the initial state, each USB and HDMI port is displayed as an access placeholder state.

5. The illegal content detection system for a display device terminal according to claim 2, characterized in that: The audio illegal content judgment unit includes a speech recognition algorithm block and a speech classification algorithm block, wherein the speech recognition algorithm block translates the sound into text and matches the keyword detection to judge the illegal content, and the speech classification algorithm block performs paralinguistic information content detection to judge whether the sound involves violence, pornography and terrorist content.

6. The illegal content detection system for a display device terminal according to claim 2, characterized in that: The text illegal content judgment unit includes a text recognition block and a text analysis block, wherein the text recognition module is used to detect whether there is illegal content in the text features, and the text analysis block is used to analyze the text semantic content and judge whether there is illegal content in the text semantics.

7. A method for using the illegal content detection system for a display device terminal according to any one of claims 1 to 6, characterized in that: The following steps are included: S1: The video stream is transmitted to the illegal multi-modal extraction module, and the features of the received video stream in three modes, namely text, picture and audio, are extracted; S2, transmitting the features extracted in S1 to the multimodal illegal content detection algorithm module, and transmitting the extracted text features, image features, and audio features to the text illegal content judgment unit, the audio illegal content judgment unit, and the video illegal content judgment unit, respectively, to judge whether there is illegal content; S3. Before the display terminal plays the video, each USB and HDMI port on the display terminal is controlled by the display screen external control device to be in the access occupied state; S4. When S2 detects that there is no illegal content in the video stream and each USB and HDMI port in S3 is not unplugged, the display terminal plays the video stream normally. Otherwise, the control relay of the display screen external control device is powered off and the display terminal stops playing.

8. The method for using the illegal content detection system for a display device terminal according to claim 7, characterized in that: S2 also includes the following steps: S21, the text violation content judgment unit first uses the text recognition block to identify whether there is any violation content in the received text features. If not, the text analysis block performs text semantic analysis on the text features to determine whether there is any violation in the text semantics; S22, the audio violation content judgment unit transmits the received audio features to the speech recognition algorithm block and the speech classification algorithm block at the same time. The speech recognition algorithm block translates the audio features into text and matches the keyword detection. The recognized text will be semantically analyzed to see if it contains sensitive violation information. The speech recognition algorithm block uses the sound classification algorithm to detect the audio features to determine whether there is violence, pornography, and terror voice. If there is, it indicates that there is a violation. S23. The video illegal content judgment unit will extract image features from frames, and judge whether the picture contains illegal content based on the image content. At the same time, it will use OCR technology to recognize the text in the picture. After matching keyword detection, the recognized text will extract text language features and then analyze it to see whether it contains sensitive information.

Images