AI-based network flow optimization platform and method

Abstract

The invention relates to the technical field of traffic optimization, in particular to an AI-based network traffic optimization platform and method. The method comprises the following steps: acquiring a monitoring video at a current moment, analyzing the monitoring video, and determining a video event; cutting and segmenting the monitoring video according to the video event to obtain a cut video; analyzing the cut video, and determining the emergency degree of the event; and determining an uploading priority according to the event emergency degree, and uploading the cut video according to the uploading priority. The bandwidth allocation is optimized, the bandwidth utilization rate is improved, and the network congestion is reduced. And the video quality is adjusted according to the uploading priority, so that the definition and real-time performance of key information are ensured. For non-key information, the video quality is reduced to save bandwidth, and information transmission is not affected. The cutting video is efficiently uploaded, and the continuity and quality of real-time monitoring are ensured.

Description

Technical Field

[0001] The present application relates to the technical field of traffic optimization, and in particular to an AI-based network traffic optimization platform and method. Background Art

[0002] As social security needs increase, video surveillance systems have been widely used. From urban security monitoring to monitoring of commercial sites and industrial areas, cameras have become an important tool for maintaining public safety and corporate management. However, the amount of video data generated by these systems is usually huge, especially today when HD or even 4K cameras are popular. This makes the transmission and storage of video data a major challenge.

[0003] In traditional video surveillance systems, uploading video recordings is usually a high-bandwidth operation. Especially when the system needs to upload a large amount of video to the cloud or central server in real time or on a scheduled basis, it may face the problem of insufficient bandwidth, resulting in upload delays, system overload, and even loss of important data. Summary of the invention

[0004] The present application provides an AI-based network traffic optimization platform and method to solve the above-mentioned problems.

[0005] In a first aspect, the present application provides an AI-based network traffic optimization method, the method comprising: Obtaining surveillance video at the current moment, analyzing the surveillance video, and determining the recorded event; According to the video recording event, the surveillance video is cut and segmented to obtain a cut video recording; Analyze the trimmed video to determine the urgency of the event; determine the upload priority according to the urgency of the event, and upload the trimmed video according to the upload priority.

[0006] Through this solution, surveillance videos can be obtained in real time to ensure the continuity and integrity of surveillance data. Through AI algorithm analysis, meaningful events can be accurately identified, the accuracy of event detection can be improved, and the false alarm rate can be reduced. Security threats, major changes and other events can be identified in a timely manner. Video clips containing major changes and other events can be cut out to reduce the upload of non-critical data and save bandwidth resources. Upload efficiency can be improved, key data can be uploaded, and the amount of data transmitted can be reduced. Emergency events can be responded to quickly to ensure the real-time transmission of critical information. Upload priority can be dynamically adjusted according to the urgency of the event to ensure that critical information is processed and transmitted first. Bandwidth allocation can be optimized to improve bandwidth utilization and reduce network congestion. Video quality can be adjusted according to upload priority to ensure the clarity and real-time nature of critical information. For non-critical information, video quality can be reduced to save bandwidth without affecting information transmission. Efficiently upload cropped videos to ensure the continuity and quality of real-time monitoring.

[0007] Optionally, analyzing the surveillance video to determine the video event includes: Analyze the surveillance video to determine whether there is a movable object; If so, the moving track and moving range of the movable object are determined according to the analysis result; Determining whether the movable object is an interference object according to the movement trajectory and the movement range; If it is an interference object, the surveillance video is processed according to the movement trajectory and the movement range to obtain a processed video without interference; The non-interference video is analyzed to determine the video event.

[0008] This solution can automatically identify movable objects such as pedestrians and vehicles in the video, reduce reliance on manual monitoring, and improve the level of automation of monitoring. By tracking the movement trajectory and range of objects, the dynamic situation of the monitoring area can be better identified. Distinguish between interfering objects such as flying insects and birds and non-interfering objects, reduce false alarms and misjudgments, and improve the accuracy of monitoring. By removing interfering objects, clearer and more accurate surveillance videos are provided, which helps to improve the accuracy and efficiency of event identification. By analyzing non-interfering videos, video events such as security threats and abnormal behaviors can be identified.

[0009] Optionally, analyzing the non-interference video to determine the video event includes: Analyze the non-interference video to determine whether there is any change in the image; If there is a change, the movement and scene change in the picture are determined based on the analysis results; Determining the picture level according to the scene change; Determine a number of analyzable events according to the movement situation and the picture level; For each analyzable event, determining the event content of the analyzable event according to the scene change situation and the movement situation; The event contents of all analyzable events are determined as the video recording events.

[0010] This solution can determine whether there are changes in the picture, quickly exclude static scenes, concentrate on processing dynamic scenes, and improve analysis efficiency. By detecting changes, specific actions and events occurring in the picture can be identified. Analyzing movement helps identify the motion state of objects in the picture. Analyzing scene changes helps identify changes in the environment. Determining the picture hierarchy helps better identify the relative positions and relationships of objects in the picture. Identifying analyzable events provides a basis for determining event content. Determining event content helps to identify and analyze events more deeply. Integrating event content forms a complete event description.

[0011] Optionally, the step of cutting and segmenting the surveillance video according to the video recording event to obtain a cut video includes: Determining, according to the screen hierarchy, an event correlation status of each analyzable event; Determine the cutting and segmentation logic according to the event association situation; The surveillance video is cut and segmented according to the movement situation, the scene change situation and the cutting and segmentation logic to obtain a cut video.

[0012] Through this solution, by analyzing the picture hierarchy, the association of each analyzable event with other events or scene elements is determined, so as to more accurately identify the background and importance of the event. According to the event association, an effective cropping and segmentation strategy is formulated to ensure that events such as major changes are fully recorded while reducing the transmission of non-critical data. Through cropping and segmentation, video clips containing only major changes and other events are extracted from the original video, reducing the amount of data and saving bandwidth resources. The cropped videos are output, which only contain the content of major changes and other events, which helps to quickly respond to and record important events, reduce unnecessary data transmission, and improve the efficiency of the monitoring system.

[0013] Optionally, analyzing the cropped video to determine the urgency of the incident includes: Determine the severity of each analyzable event according to the scene change and the movement; Determine the duration and difficulty of handling each analyzable event based on its severity; The urgency of the incident is determined based on the processing time and the processing difficulty.

[0014] This solution can identify abnormal situations in the monitoring environment by analyzing scene changes, such as sudden changes in lighting indicating an emergency. Analyzing movement helps identify the behavior patterns of objects and quickly identify security threats or abnormal behaviors. Assessing the severity of events can ensure that events such as security threats are responded to in a timely manner, while less important events can be appropriately delayed to optimize resource allocation. Estimating the processing time can help to arrange the processing sequence reasonably and ensure that emergency events can be handled quickly. Assessing the difficulty of processing can avoid wasting resources on handling simple or non-emergency events and focus more on events such as security threats. Categorizing events by urgency can dynamically adjust the processing priority according to the urgency of the event, ensuring that bandwidth resources are used most effectively.

[0015] Optionally, after uploading the cropped video according to the upload priority, the method further includes: Analyze the cropped video and the surveillance video to determine non-emergency video; Analyze the non-emergency video to determine the necessity of uploading; According to the necessity of uploading, it is determined whether to upload the non-emergency video.

[0016] This solution can reduce bandwidth usage by identifying non-emergency recordings and only upload cropped events, thereby saving network resources. Analyzing the necessity of uploads can help avoid uploading data that is not of high value or is not in real-time demand, further optimizing bandwidth usage. Choosing to upload non-emergency recordings during off-peak hours can reduce interference with real-time video transmission and improve the overall efficiency of the network. Formulating a reasonable upload strategy, including video compression and resolution adjustment, can reduce the amount of data and optimize the upload process while ensuring video quality. Formulating a reasonable upload strategy, including video compression and resolution adjustment, can reduce the amount of data and optimize the upload process while ensuring video quality. Perform upload operations to ensure that non-emergency recordings can be uploaded at the scheduled time and method to reduce interference with real-time monitoring.

[0017] Optionally, after determining whether to upload the non-emergency video according to the necessity of uploading, the method further includes: If it is determined to upload the non-emergency video, the upload task at the current moment is obtained; Analyze the upload tasks and determine the upload time and task type of each upload task; Analyze the task type and the upload time to determine the bandwidth usage during upload and the urgency of upload; Determining, according to the upload urgency, whether the upload task is the only task at the upload time; If not, analyzing the bandwidth usage to determine whether there is free time for uploading; If yes, the non-emergency video is uploaded.

[0018] Through this solution, you can identify which upload tasks need to be processed. By parsing the task details, you can identify the characteristics of each task, including the upload time and task type. Evaluate the bandwidth requirements and upload urgency of each task to provide a reference for determining task priority and bandwidth allocation. Determine which tasks need to be uploaded immediately and which can be delayed, so as to reasonably allocate bandwidth resources. Avoid uploading multiple tasks at the same time point to reduce bandwidth competition and potential conflicts. Monitor bandwidth usage in real time to provide data support for upload decisions. Determine whether there are sufficient bandwidth resources to perform non-emergency recording uploads at the current time point. If the bandwidth allows, perform non-emergency recording uploads to save storage resources and maintain monitoring continuity.

[0019] Optionally, uploading the cropped video according to the upload priority includes: Determine the upload bitrate, upload frame rate and image quality requirements according to the upload priority; According to the image quality requirement, the upload bit rate and the upload frame rate, the cropped video is adjusted to obtain a video suitable for upload, and the suitable video suitable for upload is uploaded.

[0020] This solution ensures that the cropped videos are processed and uploaded first through priority division, so as to quickly respond to emergencies and protect the safety of the monitoring targets. By setting the upload bit rate, frame rate and image quality requirements according to the priority, the bandwidth usage can be optimized, the video quality can be guaranteed, and unnecessary waste of network resources can be avoided. The bit rate, frame rate and image quality of the video can be adjusted so that the uploaded video can meet the quality requirements and adapt to the bandwidth conditions, thus improving the upload efficiency. Video compression can significantly reduce the amount of data and reduce the bandwidth usage while maintaining the availability of the video. Generate a video format suitable for uploading to ensure that the video can be correctly received and processed on the server side. Upload the video suitable for uploading to the server to ensure the integrity and real-time nature of the monitoring data.

[0021] Optionally, analyzing the bandwidth usage to determine whether there is idle time for uploading includes: Obtain historical transmission logs; analyze the historical transmission logs to determine bandwidth usage at each moment; Determine the available bandwidth at each moment according to the bandwidth occupancy; The available bandwidth and the bandwidth usage during the upload time are analyzed to determine whether there is any idle time during the upload time.

[0022] Through this solution, the usage of network bandwidth in the past period of time is collected. By analyzing historical data, the network bandwidth occupancy pattern in the past period of time can be identified. Calculating the available bandwidth at each moment helps to arrange upload tasks reasonably and avoid network congestion. Predicting the bandwidth demand generated at the upload time point provides a basis for judging whether there is enough bandwidth to perform the upload task. By comparing the available bandwidth with the bandwidth demand of the upload task, it can be determined whether the upload task is performed at the planned upload time point. That is, whether there is enough bandwidth available to perform the upload task at the upload time helps to ensure that the upload task does not affect other network activities.

[0023] In a second aspect, the present application provides an AI-based network traffic optimization platform, the platform comprising: A video analysis module, used to obtain the surveillance video at the current moment, analyze the surveillance video, and determine the video event; A video segmentation module, used for segmenting the surveillance video according to the video event to obtain a segmented video; The video uploading module is used to analyze the cropped video to determine the urgency of the event; determine the upload priority according to the urgency, and upload the cropped video according to the upload priority.

[0024] Optionally, when the video analysis module analyzes the surveillance video and determines the video event, it is used to: Analyze the surveillance video to determine whether there is a movable object; If so, determine the moving track and moving range of the movable object according to the analysis result; Determining whether the movable object is an interference object according to the movement trajectory and the movement range; If it is an interference object, the surveillance video is processed according to the movement trajectory and the movement range to obtain a processed video without interference; The non-interference video is analyzed to determine the video event.

[0025] Optionally, when the video analysis module analyzes the non-interference video and determines the video event, it is used to: Analyze the non-interference video to determine whether there are any changes in the image; If there is a change, the movement and scene change in the picture are determined based on the analysis results; Determining the picture level according to the scene change; Determine a number of analyzable events according to the movement situation and the picture level; For each analyzable event, determining the event content of the analyzable event according to the scene change situation and the movement situation; The event contents of all analyzable events are determined as the video recording events.

[0026] Optionally, the video segmentation module cuts and segments the surveillance video according to the video event, and when the cut video is obtained, is used to: Determining, according to the screen hierarchy, an event correlation status of each analyzable event; Determine the cutting and segmentation logic according to the event association situation; The surveillance video is cut and segmented according to the movement situation, the scene change situation and the cutting and segmentation logic to obtain a cut video.

[0027] Optionally, when the video uploading module analyzes the cropped video and determines the urgency of the event, it is used to: Determine the severity of each analyzable event according to the scene change and the movement; Determine the duration and difficulty of handling each analyzable event based on its severity; The urgency of the incident is determined based on the processing time and the processing difficulty.

[0028] Optionally, the AI-based network traffic optimization platform further includes a non-emergency analysis module for: Analyze the cropped video and the surveillance video to determine non-emergency video; Analyze the non-emergency video to determine the necessity of uploading; According to the necessity of uploading, it is determined whether to upload the non-emergency video.

[0029] Optionally, the AI-based network traffic optimization platform further includes a non-urgent upload module for: If it is determined to upload the non-emergency video, the upload task at the current moment is obtained; Analyze the upload tasks and determine the upload time and task type of each upload task; Analyze the task type and the upload time to determine the bandwidth usage during upload and the urgency of upload; Determining, according to the upload urgency, whether the upload task is the only task at the upload time; If not, analyzing the bandwidth usage to determine whether there is free time for uploading; If yes, the non-emergency video is uploaded.

[0030] Optionally, when uploading the cropped video according to the upload priority, the video uploading module is used to: Determine the upload bitrate, upload frame rate and image quality requirements according to the upload priority; According to the image quality requirement, the upload bit rate and the upload frame rate, the cropped video is adjusted to obtain a video suitable for upload, and the suitable video suitable for upload is uploaded.

[0031] Optionally, when the non-urgent uploading module analyzes the bandwidth usage and determines whether there is idle time for uploading, it is used to: Obtain historical transmission logs; analyze the historical transmission logs to determine bandwidth usage at each moment; Determine the available bandwidth at each moment according to the bandwidth occupancy; The available bandwidth and the bandwidth usage during the upload time are analyzed to determine whether there is any idle time during the upload time. BRIEF DESCRIPTION OF THE DRAWINGS

[0032] In order to more clearly illustrate the embodiments of the present application or the technical solutions in the prior art, 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.

[0033] Figure 1 A schematic diagram of an application scenario provided for an embodiment of the present application; Figure 2 A flowchart of an AI-based network traffic optimization method provided in one embodiment of the present application; Figure 3 A schematic diagram of the structure of an AI-based network traffic optimization platform provided in one embodiment of the present application. DETAILED DESCRIPTION

[0034] In order to make the purpose, technical scheme and advantages of the embodiments of the present application clearer, the technical scheme 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.

[0035] In addition, the term "and / or" in this article is only a description of the association relationship of associated objects, indicating that there can be three relationships. For example, A and / or B can represent: A exists alone, A and B exist at the same time, and B exists alone. In addition, the character " / " in this article, unless otherwise specified, generally means that the associated objects before and after are in an "or" relationship.

[0036] The embodiments of the present application are further described in detail below in conjunction with the drawings in the specification.

[0037] In traditional video surveillance systems, uploading video recordings is usually a high-bandwidth operation. Especially when the system needs to upload a large amount of video to the cloud or central server in real time or on a scheduled basis, it may face the problem of insufficient bandwidth, resulting in upload delays, system overload, and even loss of important data.

[0038] Based on this, the present application provides an AI-based network traffic optimization platform and method, which obtains the surveillance video at the current moment, analyzes the surveillance video, and determines the video event; according to the video event, the surveillance video is cut and segmented to obtain the cut video; the cut video is analyzed to determine the urgency of the event; and according to the urgency of the event, the upload priority is determined, and the cut video is uploaded according to the upload priority. Real-time acquisition of surveillance video ensures the continuity and integrity of surveillance data. Through AI algorithm analysis, meaningful events can be accurately identified, the accuracy of event detection can be improved, and the false alarm rate can be reduced. Security threats, major changes and other events can be identified in a timely manner. Video clips containing major changes and other events can be cut out to reduce the upload of non-critical data and save bandwidth resources. Upload efficiency can be improved, key data can be uploaded, and the amount of data transmitted can be reduced. It can respond to emergencies quickly and ensure the real-time transmission of key information. The upload priority can be dynamically adjusted according to the urgency of the event to ensure that key information is processed and transmitted first. Bandwidth allocation is optimized, bandwidth utilization is improved, and network congestion is reduced. Video quality is adjusted according to the upload priority to ensure the clarity and real-time nature of key information. For non-critical information, the video quality is reduced to save bandwidth without affecting information transmission. Efficiently upload trimmed footage to ensure the continuity and quality of real-time monitoring.

[0039] Figure 1 A schematic diagram of an application scenario provided by the present application is provided. In the uploading scenario of video surveillance data, the method provided by the present application is applied. Specifically, the method provided by the present application is applied to any server, and the server interacts with the video surveillance system to obtain the surveillance video of the video surveillance system in real time to ensure the continuity and integrity of the surveillance data. Through AI algorithm analysis, meaningful events can be accurately identified, the accuracy of event detection can be improved, and the false alarm rate can be reduced. Security threats, major changes and other events can be identified in a timely manner. Video clips containing major changes and other events can be cut out to reduce the upload of non-critical data and save bandwidth resources. Upload efficiency can be improved, key data can be uploaded, and the amount of data transmitted can be reduced. Emergency events can be responded to quickly to ensure the real-time transmission of critical information. The upload priority can be dynamically adjusted according to the urgency of the event to ensure that critical information is processed and transmitted first. Bandwidth allocation can be optimized, bandwidth utilization can be improved, and network congestion can be reduced. Video quality can be adjusted according to the upload priority to ensure the clarity and real-time nature of critical information. For non-critical information, the video quality can be reduced to save bandwidth without affecting information transmission. Efficiently upload the cut video to ensure the continuity and quality of real-time monitoring. The specific implementation method can refer to the following embodiments.

[0040] Figure 2 This is a flow chart of an AI-based network traffic optimization method provided in an embodiment of the present application. The method of this embodiment can be applied to the server in the above scenario. Figure 2 As shown, the method includes: S201, obtaining the current surveillance video, analyzing the surveillance video, and determining the video event; The current moment may be an instant time point for performing video analysis and processing.

[0041] Surveillance video can be real-time video or video data stream or video clip captured by a surveillance camera.

[0042] Specifically, the surveillance video data is obtained from the surveillance camera in real time. The surveillance video is processed by adjusting the resolution and frame rate. Image processing technology is used to extract features such as color, shape, and motion trajectory in the surveillance video. Based on the above features, an event detection algorithm is applied to identify the video events in the video.

[0043] S202, cutting and segmenting the surveillance video according to the video event to obtain a cut video; Video events can be meaningful activities or changes that occur in surveillance video.

[0044] Specifically, the surveillance video is analyzed frame by frame using an AI algorithm to identify and mark the events that occurred. The cropping points are determined based on the start and end frames of the identified events. The surveillance video is cropped into segments containing a single event or a sequence of related events, thereby obtaining a cropped video.

[0045] S203, analyzing the trimmed video to determine the urgency of the event; determining the upload priority according to the urgency of the event, and uploading the trimmed video according to the upload priority.

[0046] The cropped video may be a video segment corresponding to the identified video event extracted from the original surveillance video.

[0047] The event urgency may be a value obtained by quantifying the importance and urgency of the recorded event.

[0048] The upload priority may be determined based on the urgency of the event to determine the order in which the video clips are uploaded on the network.

[0049] Specifically, read the event type, timestamp and other information of the cropped video. Perform image quality assessment and event feature analysis on the content of the cropped video. Evaluate the urgency of the event based on the AI ​​algorithm. Classify events into high, medium and low urgency levels. Assign upload priorities to cropped videos based on the urgency of the event: for example, high-urgency events will be assigned high-priority uploads. Consider network conditions and other tasks being uploaded, and adjust the upload priority to ensure that critical tasks are executed first. Put the cropped video into the upload queue according to priority. Dynamically schedule upload tasks based on network bandwidth and the priority of the upload queue. Monitor the upload process to ensure the integrity and timeliness of data transmission.

[0050] Through this solution, surveillance videos can be obtained in real time to ensure the continuity and integrity of surveillance data. Through AI algorithm analysis, meaningful events can be accurately identified, the accuracy of event detection can be improved, and the false alarm rate can be reduced. Security threats, major changes and other events can be identified in a timely manner. Video clips containing major changes and other events can be cut out to reduce the upload of non-critical data and save bandwidth resources. Upload efficiency can be improved, key data can be uploaded, and the amount of data transmitted can be reduced. Emergency events can be responded to quickly to ensure the real-time transmission of critical information. Upload priority can be dynamically adjusted according to the urgency of the event to ensure that critical information is processed and transmitted first. Bandwidth allocation can be optimized to improve bandwidth utilization and reduce network congestion. Video quality can be adjusted according to upload priority to ensure the clarity and real-time nature of critical information. For non-critical information, video quality can be reduced to save bandwidth without affecting information transmission. Efficiently upload cropped videos to ensure the continuity and quality of real-time monitoring.

[0051] Optionally, analyze the surveillance video to determine whether there is a movable object; if so, determine the moving trajectory and moving range of the movable object based on the analysis results; determine whether the movable object is an interference object based on the moving trajectory and moving range; if it is an interference object, process the surveillance video based on the moving trajectory and moving range to obtain a processed non-interference video; analyze the non-interference video to determine the recorded event.

[0052] Movable objects can be objects such as pedestrians, vehicles, animals, etc. that can move in surveillance videos.

[0053] The analysis results can be data and information such as object detection, movement trajectory, movement range, etc. obtained by analyzing the surveillance video through AI algorithms.

[0054] The movement trajectory may be the movement path of the movable object in the surveillance video.

[0055] The moving range may be the moving area of ​​the movable object within the monitoring area.

[0056] Interference objects can be objects such as flying insects, birds, leaves, etc. that appear in the surveillance video but do not need to be paid attention to or affect event recognition.

[0057] Interference-free video can be surveillance video that has been processed to remove interference.

[0058] Specifically, read the surveillance video data and perform preprocessing such as video frame extraction and timestamp synchronization. Use image processing algorithms to detect moving objects in the video. Analyze the processed video frames and mark the detected moving objects. Track the appearance and disappearance of moving objects through continuous frame analysis to determine their existence. Calculate the bounding box of the moving object and track its movement trajectory in the video. Determine the movement range of the moving object based on the space occupied in the video frame. Based on the size, speed, movement pattern and other characteristics of the moving object and the preset interference standard, combine the machine learning classifier to assist in determining whether the moving object is an interference. If it is an interference, use the AI ​​algorithm to remove it from the video or replace it with a static background. Adjust the resolution or frame rate of the video to optimize the quality of the processed video. Perform event analysis on the processed interference-free video and use the AI ​​algorithm to identify key events and patterns. Determine the recorded event based on event characteristics such as the time of occurrence, duration, and behavior pattern.

[0059] This solution can automatically identify movable objects such as pedestrians and vehicles in the video, reduce reliance on manual monitoring, and improve the level of automation of monitoring. By tracking the movement trajectory and range of objects, the dynamic situation of the monitoring area can be better identified. Distinguish between interfering objects such as flying insects and birds and non-interfering objects, reduce false alarms and misjudgments, and improve the accuracy of monitoring. By removing interfering objects, clearer and more accurate surveillance videos are provided, which helps to improve the accuracy and efficiency of event identification. By analyzing non-interfering videos, video events such as security threats and abnormal behaviors can be identified.

[0060] Optionally, analyze the interference-free video to determine whether there are any changes in the picture; if there are any changes, determine the movement and scene changes in the picture based on the analysis results; determine the picture hierarchy based on the scene changes; determine a number of analyzable events based on the movement and picture hierarchy; for each analyzable event, determine the event content of the analyzable event based on the scene changes and movement; determine the event content of all analyzable events as recorded events.

[0061] The changes may be changes such as object movement, shape change, color change, etc. in the image content of the surveillance video.

[0062] The analysis results can be data and conclusions such as identification of picture changes, tracking of moving objects, classification of event types, etc. obtained by analyzing surveillance videos.

[0063] The movement condition may be movement characteristics such as speed, direction, and trajectory of an object in a surveillance video.

[0064] The scene change condition may be an overall change in the light intensity, weather conditions, background layout, etc. of the monitoring scene.

[0065] The picture hierarchy can be different hierarchical structures such as foreground, middle ground and background in the monitoring picture.

[0066] Analyzable events can be events that can be identified and analyzed in surveillance footage.

[0067] The event content may include specific details such as the event type, occurrence time, objects involved, and event impact of the analyzable event.

[0068] Specifically, read the interference-free video data and perform preliminary video frame processing. Use image processing technology to analyze video frames and detect changes in the picture. Identify any changes such as object movement, light changes, or scene layout changes in the picture. Determine the type and degree of change by comparing the differences between consecutive frames. Track the trajectory of moving objects in the picture and analyze their movement patterns. Identify scene changes such as the appearance or disappearance of objects and changes in scene layout. Establish the hierarchical structure of the picture based on the size, position, and occlusion relationship of objects in the picture. Use deep learning or image segmentation technology to distinguish between foreground and background. Based on the movement and picture hierarchy, identify events such as object movement and human behavior that can be further analyzed. Use event detection algorithms to mark these analyzable events. For each analyzable event, analyze its scene changes and movement to determine the specific content of the event. Use machine learning classifiers or rule engines to classify events. Based on the content of the event, it is determined to be a recorded event, and relevant information such as event type, timestamp, duration, etc. is recorded.

[0069] This solution can determine whether there are changes in the picture, quickly exclude static scenes, concentrate on processing dynamic scenes, and improve analysis efficiency. By detecting changes, specific actions and events occurring in the picture can be identified. Analyzing movement helps identify the motion state of objects in the picture. Analyzing scene changes helps identify changes in the environment. Determining the picture hierarchy helps better identify the relative positions and relationships of objects in the picture. Identifying analyzable events provides a basis for determining event content. Determining event content helps to identify and analyze events more deeply. Integrating event content forms a complete event description.

[0070] Optionally, according to the picture hierarchy, determine the event correlation of each analyzable event; determine the cutting and segmentation logic according to the event correlation; according to the movement, scene change and cutting and segmentation logic, cut and segment the surveillance video to obtain a cut video.

[0071] The event association situation may be the relationship and connection between various events in the surveillance video.

[0072] The cropping and segmentation logic may be to determine the rules and strategies for cropping and segmenting the surveillance video according to the event association situation and analysis results.

[0073] Specifically, analyze the video frames, identify the foreground objects and background, and establish the hierarchical structure of the picture. Distinguish objects of different levels according to the size, position, motion state and other characteristics of the objects. Analyze each analyzable event and identify the temporal and spatial relationships between events such as the sequence and interaction. Use association rules or graph theory models to determine the correlation between events. Based on the event association, define the basic logic of cutting and segmentation, and whether it is necessary to merge and cut related events. Determine the cutting boundaries to ensure that the cut video clips contain complete event information. Analyze the movement and scene changes during the event to determine the dynamic factors that need to be considered when cutting. Evaluate the impact of scene changes on the occlusion and light changes of the cut and segmented objects. Use video editing tools or algorithms to cut and segment the surveillance video according to the cutting and segmentation logic. Ensure that the cut video clips contain the start and end of the event, as well as key actions and changes. Save the cut video clips as independent cut videos.

[0074] Through this solution, by analyzing the picture hierarchy, the association of each analyzable event with other events or scene elements is determined, so as to more accurately identify the background and importance of the event. According to the event association, an effective cropping and segmentation strategy is formulated to ensure that events such as major changes are fully recorded while reducing the transmission of non-critical data. Through cropping and segmentation, video clips containing only major changes and other events are extracted from the original video, reducing the amount of data and saving bandwidth resources. The cropped videos are output, which only contain the content of major changes and other events, which helps to quickly respond to and record important events, reduce unnecessary data transmission, and improve the efficiency of the monitoring system.

[0075] Optionally, the severity of each analyzable event is determined based on scene changes and movement conditions; the processing time and difficulty of each analyzable event are determined based on the severity of the event; and the urgency of the event is determined based on the processing time and difficulty of the event.

[0076] The severity level may be the degree of impact of an event identified in a surveillance video on the surveillance environment or security.

[0077] Processing time can be the length of time it takes to process and respond to an event.

[0078] Processing difficulty can be the technical difficulty and resource consumption required to handle an event.

[0079] Specifically, identify and record scene changes such as light changes, appearance or disappearance of objects in the video. Track and analyze the speed, direction, and pattern of movement of objects in the video. Evaluate the potential impact of each analyzable event, such as security threats and violations, based on scene changes and movement. Use preset standards or machine learning models to quantify the severity of the event. Estimate the time required to process each analyzable event based on the complexity of the event and the required depth of analysis. Consider the current workload and predict the processing time. Analyze the number of objects, scene complexity, and other characteristics of the event to assess the difficulty of processing. Combine algorithm capabilities and historical processing data to determine the level of processing difficulty. Comprehensively evaluate the urgency of each analyzable event based on its severity, processing time, and processing difficulty. Use a priority matrix or decision tree model to classify events into high, medium, and low urgency levels.

[0080] This solution can identify abnormal situations in the monitoring environment by analyzing scene changes, such as sudden changes in lighting indicating an emergency. Analyzing movement helps identify the behavior patterns of objects and quickly identify security threats or abnormal behaviors. Assessing the severity of events can ensure that events such as security threats are responded to in a timely manner, while less important events can be appropriately delayed to optimize resource allocation. Estimating the processing time can help to arrange the processing sequence reasonably and ensure that emergency events can be handled quickly. Assessing the difficulty of processing can avoid wasting resources on handling simple or non-emergency events and focus more on events such as security threats. Categorizing events by urgency can dynamically adjust the processing priority according to the urgency of the event, ensuring that bandwidth resources are used most effectively.

[0081] Optionally, analyze the cropped video and the surveillance video to determine the non-emergency video; analyze the non-emergency video to determine the necessity of uploading; and determine whether to upload the non-emergency video based on the necessity of uploading.

[0082] Non-emergency footage can be footage of surveillance footage containing events or scene changes that are not of sufficient magnitude to warrant immediate attention or require real-time processing.

[0083] The necessity of uploading can be determined based on monitoring requirements, the importance and real-time nature of the video content, and the availability of network resources to determine whether the video data needs to be uploaded to the server.

[0084] Specifically, read the cropped video and surveillance video, and extract necessary information such as event type, timestamp, event description, etc. Use event analysis algorithms such as machine learning models or rule engines to classify events in the video. Classify events according to their urgency and identify non-emergency events and corresponding videos. Combine expert knowledge and historical data analysis to more accurately determine non-emergency videos. Consider the content of non-emergency videos and evaluate their importance for monitoring purposes and whether they contain key information or evidence. Analyze storage costs, bandwidth resources, and other system limitations to determine the feasibility of uploading non-emergency videos. Based on the analysis results of the necessity of upload, decide whether to add non-emergency videos to the upload queue.

[0085] This solution can reduce bandwidth usage by identifying non-emergency recordings and only upload cropped events, thereby saving network resources. Analyzing the necessity of uploads can help avoid uploading data that is not of high value or is not in real-time demand, further optimizing bandwidth usage. Choosing to upload non-emergency recordings during off-peak hours can reduce interference with real-time video transmission and improve the overall efficiency of the network. Formulating a reasonable upload strategy, including video compression and resolution adjustment, can reduce the amount of data and optimize the upload process while ensuring video quality. Formulating a reasonable upload strategy, including video compression and resolution adjustment, can reduce the amount of data and optimize the upload process while ensuring video quality. Perform upload operations to ensure that non-emergency recordings can be uploaded at the scheduled time and method to reduce interference with real-time monitoring.

[0086] Optionally, if it is determined to upload non-emergency videos, obtain the upload tasks at the current moment; parse the upload tasks to determine the upload time and task type of each upload task; analyze the task type and upload time to determine the bandwidth usage during upload and the upload urgency; based on the upload urgency, determine whether the upload task is the only task at the upload time; if not, analyze the bandwidth usage to determine whether there is free time for the upload; if so, upload the non-emergency videos.

[0087] An upload task can be a file or data that needs to be transferred to the server through the network.

[0088] The upload time may be the time point when the upload task is scheduled or actually executed.

[0089] The task type may be a category or nature of an upload task such as urgent upload, secondary upload, or non-urgent upload.

[0090] The bandwidth usage may be the network bandwidth resources occupied by the upload task during the transmission process.

[0091] Upload urgency may be the importance and urgency of the upload task.

[0092] A unique task can be one that has no other tasks competing for network resources at a certain point in time.

[0093] Idle means that the network bandwidth is not fully occupied and there are remaining resources that can be used by other tasks.

[0094] Specifically, retrieve the list of currently pending upload tasks from the upload queue or task management system. Obtain detailed information such as the ID, file size, and estimated upload time of each task. Analyze each upload task to determine its upload time and task type such as urgent task or regular task. Prioritize tasks based on task type and upload time. Evaluate bandwidth requirements based on task type. Analyze the estimated bandwidth usage in a specific time period in combination with the upload time. Calculate the bandwidth usage of each upload task in a specific time period. Evaluate the urgency of the upload based on the importance and deadline of the task. Check whether there are other tasks waiting in line for upload at the current upload time. If it is the only task, the upload can be scheduled directly; if not, further analysis is required. Evaluate the current network bandwidth usage, including occupied bandwidth and available bandwidth. If there is free bandwidth, the upload of non-urgent recordings can be scheduled. If it is determined that non-urgent recordings can be uploaded, add them to the upload queue.

[0095] Through this solution, you can identify which upload tasks need to be processed. By parsing the task details, you can identify the characteristics of each task, including the upload time and task type. Evaluate the bandwidth requirements and upload urgency of each task to provide a reference for determining task priority and bandwidth allocation. Determine which tasks need to be uploaded immediately and which can be delayed, so as to reasonably allocate bandwidth resources. Avoid uploading multiple tasks at the same time point to reduce bandwidth competition and potential conflicts. Monitor bandwidth usage in real time to provide data support for upload decisions. Determine whether there are sufficient bandwidth resources to perform non-emergency recording uploads at the current time point. If the bandwidth allows, perform non-emergency recording uploads to save storage resources and maintain monitoring continuity.

[0096] Optionally, the upload bit rate, upload frame rate and image quality requirements are determined based on the upload priority; based on the image quality requirements, upload bit rate and upload frame rate, the video is adjusted and cropped to obtain a suitable upload video, and the suitable upload video is uploaded.

[0097] The upload bitrate can be the amount of data transmitted per second during the video upload process.

[0098] The upload frame rate can be the number of frames transmitted per second during the video upload process.

[0099] The image quality requirement may be the image quality standard that the uploaded video needs to meet.

[0100] The suitable upload video may be a video file that has been adjusted and compressed and is suitable for uploading under current network conditions.

[0101] Specifically, assign a priority to each upload task based on the urgency, importance, and other business rules of the event. Ensure that high-priority tasks are given priority in resource allocation and upload scheduling. Determine the appropriate upload bitrate based on the priority of the upload task. High-priority tasks require a higher bitrate to maintain video quality. Determine the upload frame rate. High-priority tasks require a higher frame rate to maintain video smoothness. Determine image quality requirements such as resolution and color depth based on video content and viewing requirements. Use video processing software or algorithms to adjust the bitrate of the cropped video to adapt to the determined upload bitrate. Adjust the frame rate of the cropped video to match the determined upload frame rate. Based on image quality requirements, the cropped video needs to be re-encoded to optimize video quality. Save the adjusted cropped video as a new file, which will be used as the video suitable for upload. Generate metadata for information such as video parameters, encoding format, and file size for the video suitable for upload. Add the video suitable for upload to the upload queue and arrange the upload order according to priority.

[0102] This solution ensures that the cropped videos are processed and uploaded first through priority division, so as to quickly respond to emergencies and protect the safety of the monitoring targets. By setting the upload bit rate, frame rate and image quality requirements according to the priority, the bandwidth usage can be optimized, the video quality can be guaranteed, and unnecessary waste of network resources can be avoided. The bit rate, frame rate and image quality of the video can be adjusted so that the uploaded video can meet the quality requirements and adapt to the bandwidth conditions, thus improving the upload efficiency. Video compression can significantly reduce the amount of data and reduce the bandwidth usage while maintaining the availability of the video. Generate a video format suitable for uploading to ensure that the video can be correctly received and processed on the server side. Upload the video suitable for uploading to the server to ensure the integrity and real-time nature of the monitoring data.

[0103] Optionally, obtain historical transmission logs; analyze historical transmission logs to determine bandwidth usage at each moment; determine available bandwidth at each moment based on bandwidth usage; analyze available bandwidth and bandwidth usage at upload time to determine whether there is idle time at upload time.

[0104] The historical transmission log may be a log file that records the network data transmission status in the past period of time.

[0105] Bandwidth usage refers to the usage of network bandwidth within a certain period of time.

[0106] Available bandwidth refers to the unused bandwidth resources in the network within a certain period of time.

[0107] Specifically, collect historical transmission log files from network devices and servers. Ensure the integrity and accuracy of log files, and verify the source and format of log files. Use log analysis tools or custom scripts to parse log files and extract transmission data. Identify key information such as transmission time, transmission speed, and file size in the log. Calculate the bandwidth utilization rate at each moment based on historical transmission data. Add up the bandwidth utilization of different transmission tasks to obtain the total bandwidth utilization. Calculate the available bandwidth at each moment based on the maximum bandwidth capacity of network devices and historical bandwidth utilization. Consider network peak and valley periods, as well as any planned network maintenance or upgrades. Predict the bandwidth requirements of the upcoming upload tasks, such as upload bitrate, file size, and upload duration. Compare the predicted bandwidth requirements with the historical bandwidth utilization to evaluate the bandwidth pressure during the upload time. Based on the available bandwidth and upload requirements, determine whether there is enough bandwidth available during the planned upload time.

[0108] Through this solution, the usage of network bandwidth in the past period of time is collected. By analyzing historical data, the network bandwidth occupancy pattern in the past period of time can be identified. Calculating the available bandwidth at each moment helps to arrange upload tasks reasonably and avoid network congestion. Predicting the bandwidth demand generated at the upload time point provides a basis for judging whether there is enough bandwidth to perform the upload task. By comparing the available bandwidth with the bandwidth demand of the upload task, it can be determined whether the upload task is performed at the planned upload time point. That is, whether there is enough bandwidth available to perform the upload task at the upload time helps to ensure that the upload task does not affect other network activities.

[0109] Figure 3 A schematic diagram of the structure of an AI-based network traffic optimization platform provided in one embodiment of the present application is shown in FIG. Figure 3 As shown, the AI-based network traffic optimization platform 300 of this embodiment includes: a video analysis module 301, a video segmentation module 302, and a video upload module 303. The video analysis module 301 is used to obtain the surveillance video at the current moment, analyze the surveillance video, and determine the video event; The video segmentation module 302 is used to segment the surveillance video according to the video event to obtain a segmented video; The video uploading module 303 is used to analyze the cropped video to determine the urgency of the event; determine the upload priority according to the urgency, and upload the cropped video according to the upload priority.

[0110] Optionally, when the video analysis module 301 analyzes the surveillance video and determines the video event, it is used to: Analyze the surveillance video to determine whether there is a movable object; If so, the moving track and moving range of the movable object are determined according to the analysis result; Determining whether the movable object is an interference object according to the movement trajectory and the movement range; If it is an interference object, the surveillance video is processed according to the movement trajectory and the movement range to obtain a processed video without interference; The non-interference video is analyzed to determine the video event.

[0111] Optionally, when the video analysis module 301 analyzes the non-interference video and determines the video event, it is used to: Analyze the non-interference video to determine whether there is any change in the image; If there is a change, the movement and scene change in the picture are determined based on the analysis results; Determining the picture level according to the scene change; Determine a number of analyzable events according to the movement situation and the picture level; For each analyzable event, determining the event content of the analyzable event according to the scene change situation and the movement situation; The event contents of all analyzable events are determined as the video recording events.

[0112] Optionally, the video segmentation module 302 segments the surveillance video according to the video event, and when the segmented video is obtained, is used to: Determining, according to the screen hierarchy, an event correlation status of each analyzable event; Determine the cutting and segmentation logic according to the event association situation; The surveillance video is cut and segmented according to the movement situation, the scene change situation and the cutting and segmentation logic to obtain a cut video.

[0113] Optionally, when the video uploading module 303 analyzes the cropped video and determines the urgency of the event, it is used to: Determine the severity of each analyzable event according to the scene change and the movement; Determine the duration and difficulty of handling each analyzable event based on its severity; The urgency of the incident is determined based on the processing time and the processing difficulty.

[0114] Optionally, the AI-based network traffic optimization platform 300 further includes a non-emergency analysis module 304, which is used to: Analyze the cropped video and the surveillance video to determine non-emergency video; Analyze the non-emergency video to determine the necessity of uploading; According to the necessity of uploading, it is determined whether to upload the non-emergency video.

[0115] Optionally, the AI-based network traffic optimization platform 300 further includes a non-urgent upload module 305, which is used to: If it is determined to upload the non-emergency video, the upload task at the current moment is obtained; Analyze the upload tasks and determine the upload time and task type of each upload task; Analyze the task type and the upload time to determine the bandwidth usage during upload and the urgency of upload; Determining, according to the upload urgency, whether the upload task is the only task at the upload time; If not, analyzing the bandwidth usage to determine whether there is free time for uploading; If yes, the non-emergency video is uploaded.

[0116] Optionally, when uploading the cropped video according to the upload priority, the video uploading module 303 is used to: Determine the upload bitrate, upload frame rate and image quality requirements according to the upload priority; According to the image quality requirement, the upload bit rate and the upload frame rate, the cropped video is adjusted to obtain a video suitable for upload, and the suitable video suitable for upload is uploaded.

[0117] Optionally, when the non-urgent uploading module 305 analyzes the bandwidth usage and determines whether there is idle time for uploading, it is used to: Obtain historical transmission logs; analyze the historical transmission logs to determine bandwidth usage at each moment; Determine the available bandwidth at each moment according to the bandwidth occupancy; The available bandwidth and the bandwidth usage during the upload time are analyzed to determine whether there is any idle time during the upload time.

[0118] The platform of this embodiment can be used to execute the method of any of the above embodiments, and its implementation principles and technical effects are similar and will not be repeated here.

Claims

1. A network traffic optimization method based on AI, characterized in that: include: Obtaining surveillance video at the current moment, analyzing the surveillance video, and determining the recorded event; According to the video recording event, the surveillance video is cut and segmented to obtain a cut video recording; Analyze the cropped video to determine the urgency of the incident; And according to the urgency of the event, an upload priority is determined, and the cropped video is uploaded according to the upload priority.

2. The method according to claim 1, characterized in that: The step of analyzing the surveillance video and determining the video event includes: Analyze the surveillance video to determine whether there is a movable object; If so, the moving track and moving range of the movable object are determined according to the analysis result; Determining whether the movable object is an interference object according to the movement trajectory and the movement range; If it is an interference object, the surveillance video is processed according to the movement trajectory and the movement range to obtain a processed video without interference; The non-interference video is analyzed to determine the video event.

3. The method according to claim 2, characterized in that The analyzing the non-interference video to determine the video event includes: Analyze the non-interference video to determine whether there is any change in the image; If there is a change, the movement and scene change in the picture are determined based on the analysis results; Determining the picture level according to the scene change; Determine a number of analyzable events according to the movement situation and the picture level; For each analyzable event, determining the event content of the analyzable event according to the scene change situation and the movement situation; The event contents of all analyzable events are determined as the video recording events.

4. The method according to claim 3, characterized in that The step of cutting and segmenting the surveillance video according to the video recording event to obtain a cut video recording includes: Determining, according to the screen hierarchy, an event correlation status of each analyzable event; Determine the cutting and segmentation logic according to the event association situation; The surveillance video is cut and segmented according to the movement situation, the scene change situation and the cutting and segmentation logic to obtain a cut video.

5. The method according to claim 3, characterized in that: The analyzing the cropped video to determine the urgency of the incident includes: Determine the severity of each analyzable event according to the scene change and the movement; Determine the duration and difficulty of handling each analyzable event based on its severity; The urgency of the incident is determined based on the processing time and the processing difficulty.

6. The method according to claim 1, characterized in that After uploading the cropped video according to the upload priority, the method further includes: Analyze the cropped video and the surveillance video to determine non-emergency video; Analyze the non-emergency video to determine the necessity of uploading; According to the necessity of uploading, it is determined whether to upload the non-emergency video.

7. The method according to claim 6, characterized in that After determining whether to upload the non-emergency video according to the necessity of uploading, the method further includes: If it is determined to upload the non-emergency video, the upload task at the current moment is obtained; Analyze the upload tasks and determine the upload time and task type of each upload task; Analyze the task type and the upload time to determine the bandwidth usage during upload and the urgency of upload; Determining, according to the upload urgency, whether the upload task is the only task at the upload time; If not, analyzing the bandwidth usage to determine whether there is free time for uploading; If yes, the non-emergency video is uploaded.

8. The method according to claim 1, characterized in that The uploading of the cropped video according to the upload priority includes: Determine the upload bitrate, upload frame rate and image quality requirements according to the upload priority; According to the image quality requirement, the upload bit rate and the upload frame rate, the cropped video is adjusted to obtain a video suitable for upload, and the suitable video suitable for upload is uploaded.

9. The method according to claim 7, characterized in that: The analyzing the bandwidth usage to determine whether there is free time for uploading includes: Obtain historical transmission logs; analyze the historical transmission logs to determine bandwidth usage at each moment; Determine the available bandwidth at each moment according to the bandwidth occupancy; The available bandwidth and the bandwidth usage during the upload time are analyzed to determine whether there is any idle time during the upload time.

10. An AI-based network traffic optimization platform, characterized in that: include: A video analysis module, used to obtain the surveillance video at the current moment, analyze the surveillance video, and determine the video event; A video segmentation module, used for segmenting the surveillance video according to the video event to obtain a segmented video; The video uploading module is used to analyze the cropped video to determine the urgency of the event; determine the upload priority according to the urgency, and upload the cropped video according to the upload priority.

Images