Video detection method, electronic equipment and computer readable storage medium

Abstract

The invention discloses a video detection method, electronic equipment and a computer readable storage medium device, and the method comprises the steps that target edge node equipment analyzes a target video frame to obtain an initial analysis result of the target video frame, and the target video frame is reported to the target edge node equipment by video collection equipment; obtaining a historical analysis result of at least one historical video frame; wherein the target video frame and the at least one historical video frame are from the same target video, the analysis result of each video frame of the target video is obtained by at least analyzing and executing two pieces of edge node equipment, and the storage positions of the analysis results of each video frame of the target video are the same; and performing static target filtering on the initial analysis result by using the historical analysis result of the at least one historical video frame to obtain a target analysis result. In this way, the video can be effectively detected.

Description

Technical Field

[0001] The present application relates to the technical field of video detection, and in particular to a video detection method, an electronic device and a computer-readable storage medium. Background Art

[0002] At present, video IoT devices have been widely used in many scenarios (such as home scenarios, work scenarios, etc.). In addition, intelligent detection of videos collected by IoT devices has also been widely used.

[0003] However, there is currently no effective post-processing strategy for the detection results obtained by intelligent detection of videos, resulting in a poor user experience. Summary of the invention

[0004] The main technical problem solved by the present application is to provide a video detection method, an electronic device and a computer-readable storage medium, which can effectively detect the video.

[0005] To solve the above technical problems, a technical solution adopted in the present application is: providing a video detection method, the method comprising: a target edge node device analyzes a target video frame to obtain an initial analysis result of the target video frame, wherein the target video frame is reported to the target edge node device by a video acquisition device; and obtaining a historical analysis result of at least one historical video frame; wherein the target video frame and at least one historical video frame are derived from the same target video, the analysis results of each video frame of the target video are obtained by performing analysis and execution by at least two edge node devices, and the storage location of the analysis results of each video frame of the target video is the same; using the historical analysis result of at least one historical video frame, static target filtering is performed on the initial analysis result to obtain a target analysis result.

[0006] Among them, the initial analysis result of the target video frame includes the object information of each target object in the target video frame, the object information of the target object includes the target category of the target object and the position of the target object, the historical analysis result of the historical video frame includes the object information of each historical object in the historical video frame, and the object information of the historical object includes the historical category of the historical object and the position of the historical object; using the historical analysis result of at least one historical video frame, static target filtering is performed on the initial analysis result to obtain the target analysis result, including: for each target object in the target video frame, determining the position overlap between the target object and the historical object of the same category; based on the corresponding position overlap of the target object, determining whether the target object is a static target; from the initial analysis result, deleting the object information corresponding to each static target to obtain the target analysis result.

[0007] Among them, the characterization value of the object's position includes the object's minimum circumscribed rectangular frame; and / or, based on the position overlap corresponding to the target object, determining whether the target object is a static target, including: in response to the position overlap corresponding to the target object being greater than or equal to the overlap threshold, determining that the target object is a static target; in response to the position overlap of the target object being less than the overlap threshold, determining that the target object is a dynamic target.

[0008] The at least one historical video frame is a video frame adjacent to the target video frame in the target video.

[0009] Among them, each edge node device is used to process at least two video frames in the target video, wherein the target video frame includes the first video frame processed by the target edge node device, and the corresponding historical video frame is a video frame processed by any other edge node device; and / or, the target video frame includes a non-first video frame processed by the target edge node device, and the corresponding historical video frame is a video frame processed by the target edge node device.

[0010] The video detection method further includes: in response to the target analysis result including the analysis result of the dynamic target, feeding back the target analysis result to the user.

[0011] The target video frame is analyzed to obtain an initial analysis result of the target video frame, including: analyzing the target video frame to obtain a reference analysis result of the target video frame; and filtering the reference analysis result by using a target filtering strategy to obtain an initial analysis result.

[0012] Among them, the target filtering strategy includes filtering the reference analysis results in the target video frame that do not belong to the user's area of ​​interest; and / or, the reference analysis results of the target video frame include object information of each target object in the target video frame, the object information includes the target category of the target object and the confidence that the target object is the target category, and the target filtering strategy includes filtering the object information of the target object whose confidence is lower than the confidence threshold.

[0013] The video detection method further includes: obtaining video frames containing dynamic targets; condensing video frames corresponding to dynamic targets of the same category into target videos of corresponding categories; and feeding back target videos corresponding to dynamic targets of each category to the user.

[0014] In order to solve the above technical problems, another technical solution adopted by the present application is: to provide an electronic device, the electronic device includes a memory and a processor, the memory stores program instructions, and the program instructions can be executed to implement the above video detection method.

[0015] In order to solve the above technical problem, another technical solution adopted by the present application is: providing a computer-readable storage medium, the computer-readable storage medium stores program instructions, and the program instructions can be executed to implement the above video detection method.

[0016] In the above technical solution, the analysis results of each video frame of the target video are executed by at least two edge node devices, that is, at least two edge node devices analyze and filter each video frame of the target video (such as static target filtering, user-uninterested area filtering, confidence filtering, etc.), etc., which breaks through the filtering limitations of a single edge node device and improves the robustness and efficiency of the video detection system.

[0017] In addition, the storage location of the analysis results of each video frame of the target video is the same, that is, the analysis results corresponding to different video frames in the target video processed by different edge node devices are stored in the same storage location. Therefore, when the video frames of the target video are processed by at least two edge node devices, the static target filtering of the initial analysis results of each video frame is not affected. BRIEF DESCRIPTION OF THE DRAWINGS

[0018] Figure 1 It is a flow chart of an embodiment of a video detection method provided by the present application;

[0019] Figure 2 It is a diagram of the target video file implementation mechanism provided by this application;

[0020] Figure 3 This is a diagram of the video frame aggregation implementation mechanism provided by this application;

[0021] Figure 4 yes Figure 1 The flowchart of step S11 is shown as an embodiment;

[0022] Figure 5 yes Figure 1 The flowchart of step S13 is shown as an embodiment;

[0023] Figure 6 It is a schematic diagram of an embodiment of a minimum outer rectangular frame of an object provided by the present application;

[0024] Figure 7 It is a schematic diagram of an embodiment of position overlap provided by the present application;

[0025] Figure 8 It is a structural schematic diagram of an embodiment of a video detection device provided by the present application;

[0026] Fig. 9 It is a structural schematic diagram of an embodiment of an electronic device provided by the present application;

[0027] Fig.10 It is a structural schematic diagram of an embodiment of a computer-readable storage medium provided by the present application. DETAILED DESCRIPTION

[0028] The scheme of the embodiment of the present application is described in detail below in conjunction with the drawings of the specification.

[0029] In the following description, for the purpose of explanation rather than limitation, specific details such as specific system structures, interfaces, and technologies are provided to facilitate a thorough understanding of the present application.

[0030] The term "and / or" in this article is only a description of the association relationship of associated objects, indicating that there may 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 generally indicates that the objects associated before and after are in an "or" relationship. In addition, "many" in this article means two or more than two. In addition, the term "at least one" in this article means any combination of at least two of any one or more of a plurality of, for example, including at least one of A, B, and C, can mean including any one or more elements selected from the set consisting of A, B, and C.

[0031] See also Figure 1 , Figure 1 is a flow chart of an embodiment of the video detection method provided by the present application. It should be noted that if there are substantially the same results, this embodiment does not Figure 1 The process sequence shown is limited. Figure 1 As shown, this embodiment includes:

[0032] Step S11: The target edge node device analyzes the target video frame to obtain an initial analysis result of the target video frame.

[0033] In this implementation, the target edge node device analyzes the target video frame to obtain an initial analysis result of the target video frame; wherein the target video frame is reported to the target edge node device by the video acquisition device.

[0034] In one embodiment, the target edge node device can directly use the reference analysis result obtained by analyzing the target video frame as the initial analysis result. Of course, in other embodiments, the target edge node device can also use the analysis result after filtering the reference analysis result as the initial analysis result, which is not limited here.

[0035] In one embodiment, the initial analysis result of the target video frame is also sent to the storage location so that the initial analysis result of the target video frame can be used as a historical video frame corresponding to other video frames to be detected to perform static target filtering of the analysis results of the video frames to be detected.

[0036] Step S12: Obtain a historical analysis result of at least one historical video frame.

[0037] In this embodiment, a historical analysis result of at least one historical video frame is obtained; wherein, the target video frame and at least one historical video frame are derived from the same target video, and the analysis result of each video frame of the target video is obtained by executing at least two edge node devices, and the storage location of the analysis result of each video frame of the target video is the same, that is, the analysis result of each video frame of the target video is stored in the same distributed cache.

[0038] The analysis results of each video frame of the target video are executed by at least two edge node devices, that is, at least two edge node devices analyze and filter each video frame of the target video (such as static target filtering, user-uninterested area filtering, confidence filtering, etc.), etc., breaking through the filtering limitations of a single edge node device and improving the robustness and efficiency of the video detection system.

[0039] Among them, there is no limitation on the number of edge node devices executed and which video frames in the target video are processed by different edge node devices, and specific settings can be made according to actual use needs. For example, 2, 3, 4, 5 or 10 edge node devices. For another example, take 3 edge node devices and the target video includes 30 video frames as an example: in one case, the first edge node device is used to process 1-10 video frames, the second edge node device is used to process 11-20 video frames, and the third edge node device is used to process 21-30 video frames; in another case, the first edge node device is used to process 1-5 video frames and 16-20 video frames, the second edge node device is used to process 6-10 video frames and 21-25 video frames, and the third edge node device is used to process 11-15 video frames and 26-30 video frames.

[0040] The storage location of the analysis results of each video frame of the target video is the same, that is, the analysis results corresponding to different video frames in the target video processed by different edge node devices are stored in the same storage location, that is, the analysis results of each video frame of the target video are stored in the same distributed cache. Since the initial analysis results of the target video frame need to be filtered using the historical analysis results of the historical video frame in the future, if the historical analysis results of the historical video frame are obtained by other edge node devices processing the historical video frame, and the analysis results corresponding to each edge node device are not stored in the same storage location, then the target edge node device cannot obtain the historical analysis results of the historical video frame, resulting in the inability to filter the initial analysis results of the target video frame. Therefore, by setting the analysis results of each video frame of the target video to be stored in the same location, that is, by setting the analysis results of each video frame of the target video to be stored in the same distributed cache, it is possible to use at least two edge node devices to process the video frames of the target video without affecting the static target filtering of the initial analysis results of each video frame.

[0041] It should be noted that the target video can be captured by home video equipment or by video equipment in the workplace or outdoors. That is, the video detection method provided in the present application can be used to detect videos captured by home video equipment, or it can be used to detect videos captured by video equipment in the workplace, outdoors, etc.

[0042] In addition, the analysis results of each video frame of the target video stored in the storage location are not limited. If the target edge node device uses the analysis result after filtering the reference analysis result as the initial analysis result, then the analysis result of each video frame of the target video stored in the storage location is the analysis result after filtering the reference analysis result. If the target edge node device uses the reference analysis result obtained by analyzing the target video frame directly as the initial analysis result, then the analysis result of each video frame of the target video stored in the storage location is the reference analysis result.

[0043] For example, the target filtering strategy is to filter the reference analysis results that do not belong to the user's area of ​​interest in the target video frame, then the analysis results of each video frame of the target video stored in the storage location are the analysis results of the user's area of ​​interest.

[0044] Step S13: Using the historical analysis result of at least one historical video frame, static target filtering is performed on the initial analysis result to obtain a target analysis result.

[0045] In this embodiment, the initial analysis result is filtered for static targets using the historical analysis result of at least one historical video frame to obtain the target analysis result. The static target in the target video frame can be determined by combining the initial analysis result of the current video frame with the historical analysis results of the historical video frames.

[0046] If the target object is a static target, and the position of the video device that captures the target video does not change substantially, then the position of the target object in the historical video frame and the position in the target video frame are substantially unchanged. Therefore, in one embodiment, the position overlap of the target object in the historical video frame and the target video frame can be used to determine whether the target object is a static target.

[0047] In one embodiment, at least one historical video frame is a video frame adjacent to the target video frame in the target video. In other words, there is one historical video frame, specifically a video frame adjacent to the target video frame.

[0048] In a specific implementation, each edge node device is used to process at least two video frames in the target video, wherein the target video frame includes the first video frame processed by the target edge node device, that is, including multiple video frames reported for the first time and subsequent interval time frame sampling, and the corresponding historical video frame is a video frame processed by any other edge node device. Since the target video includes multiple video frames, and the multiple video frames are processed by at least two edge node devices, and the historical video frame is one, specifically a video frame adjacent to the target video frame; therefore, when the target video frame is the first processed video frame of the target edge node device, the historical video frame must be processed by any other edge node device.

[0049] However, it should be noted that the storage location of the analysis results of each video frame of the target video is the same, that is, the analysis results of each video frame of the target video are stored in the same distributed cache. Therefore, the target edge node device can obtain the historical analysis results of at least one historical video frame from the storage location, which does not affect the static target filtering of the initial analysis results of the target video frame.

[0050] In other specific implementations, each edge node device is used to process at least two video frames in the target video, wherein the target video frame includes multiple non-first video frames processed by the target edge node device, and the corresponding historical video frame is a video frame processed by the target edge node device. Since there is only one historical video frame, specifically a video frame adjacent to the target video frame, when the target video frame is a non-first processed video frame of the target edge node device, the historical video frame must be processed by the target edge node device itself.

[0051] In one embodiment, the target analysis results are also fed back to the user. In the video monitoring scenario, a large amount of data is generated, and the analysis results of static targets may be unimportant or repetitive to the user. Therefore, feeding back the analysis results of dynamic targets to the user can, on the one hand, reduce the redundant information that the user needs to view and improve the efficiency and pertinence of monitoring; on the other hand, dynamic targets are often associated with abnormal or important events, which allows users to identify and respond to key events more quickly.

[0052] In a specific implementation, the target analysis result is fed back to the user, specifically: in response to the target analysis result including the analysis result of the dynamic target, the target analysis result is fed back to the user. That is, the target analysis result is fed back to the user only when there is a relevant analysis result of the dynamic target.

[0053] In one embodiment, video frames with dynamic targets are also obtained; video frames corresponding to dynamic targets of the same category are condensed into target videos of the corresponding category; and target videos of dynamic targets of each category are fed back to the user. In other words, video frames with the same dynamic targets are condensed into short videos, so that users can spend less time to quickly browse events of interest. For example, video frames with pets are condensed into short videos, so that users can spend less time to quickly browse the situation of the pets.

[0054] Specifically, the following steps are included: Step 1: Extract video frame intelligent event detection, cache video frame index and detection results; Step 2: Download and transcode video frames according to the cached frames of the video frame detection results to form a new target video file. Step 3: Store the new target video file and m3u8 video index records in the library and upload them to the designated cloud storage node device.

[0055] In a specific implementation, the user can view the target video, download the target video and play it on the terminal device.

[0056] It should be noted that if Figure 2 As shown, Figure 2 This is a diagram of the target video file implementation mechanism provided by this application. The m3u8 file aggregates video frames of the same type of target into short videos. Users can query the m3u8 condensed video records through the terminal device, and then view, download, and play the target video file, realizing fast viewing, downloading, and playback. Among them, Figure 2 "EXT-X-TARGETDURATION" indicates the number of cover image files; Figure 2 The "EXTINF" in the text box indicates the time of the cover image. Figure 2 "EXT-X-IF" in the header indicates the frame ID, offset, and length of the frame data of the cover image file; Figure 2"EXT-X-ENDLIST" in the index file indicates the end.

[0057] Figure 3 This is a diagram of the video frame aggregation implementation mechanism provided by this application. Figure 3 The data frame details of the intelligent event video frame set file shown in the figure periodically select the intelligent event video frames generated by the device, such as all video frames in which people appear, transcode the video frames and rewrite the timestamps to form a new cloud video file.

[0058] See also Figure 4 , Figure 4 yes Figure 1 It should be noted that if there is substantially the same result, this embodiment does not use Figure 4 The process sequence shown is limited. Figure 4 As shown, this embodiment includes:

[0059] Step S41: Analyze the target video frame to obtain a reference analysis result of the target video frame.

[0060] In this implementation, the target video frame is analyzed to obtain a reference analysis result of the target video frame.

[0061] The reference analysis result of the target video frame obtained by the analysis is not limited and can be specifically set according to actual use needs. For example, the reference analysis result of the target video frame includes the object information of each target object in the target video frame, and the object information of the target object includes the target category of the target object, the confidence that the target object is the target category, and the position of the target object in the target video frame (for example, the minimum circumscribed rectangular frame of the target object can be used to represent the position of the target object in the target video frame).

[0062] Step S42: Filter the reference analysis results using the target filtering strategy to obtain initial analysis results.

[0063] In this embodiment, the reference analysis results are filtered using the target filtering strategy to obtain the initial analysis results. The reference analysis results are filtered using the target filtering strategy, which can filter out invalid analysis results and retain valid analysis results, thereby reducing the amount of data and improving video detection efficiency.

[0064] In one embodiment, the target filtering strategy includes filtering the reference analysis results that do not belong to the user's area of ​​interest in the target video frame. The target filtering strategy includes filtering the reference analysis results that do not belong to the user's area of ​​interest in the target video frame, that is, the relevant analysis results of the user's non-interested area will be filtered out, and only the relevant analysis results of the user's area of ​​interest will be retained. The target analysis results fed back to the user subsequently are the target analysis results of the user's area of ​​interest, which can feed back the video detection results to the user in a more targeted manner.

[0065] Among them, the user's area of ​​interest can be any area in the target video frame. For example, the target video frame includes a living room area, a dining room area, a balcony area and a garden area. The user's area of ​​interest can be the living room area, the dining room area, the balcony area, the garden area, etc. in the target video frame.

[0066] For example, take the target filtering strategy of filtering the reference analysis results that do not belong to the living room area in the target video frame as an example: the target video frame includes the living room area, the dining room area and the balcony area, there are target objects A and target object B in the living room area, there is a target object C in the dining room area, and there is a target object D in the balcony area; because the user's area of ​​interest is the living room area, the reference analysis results of the target object C in the dining room area and the reference analysis results of the target object D in the balcony area are filtered out; then, the initial analysis result is the relevant analysis result of the target object A and the target object B in the living room area.

[0067] In a specific implementation, the user can customize the region of interest to achieve the region of interest customization capability. For example, the target video frame includes a living room area, a dining room area, a balcony area, and a garden area, and the user can set the region of interest to be the living room area, the dining room area, the balcony area, the garden area, etc.

[0068] In other embodiments, the reference analysis result of the target video frame includes the object information of each target object in the target video frame, the object information includes the target category of the target object and the confidence that the target object is the target category, and the target filtering strategy includes filtering the object information of the target object whose confidence is lower than the confidence threshold. In other words, the analysis results with insufficient detection accuracy will be filtered out, and only the analysis results with high detection accuracy will be retained. The target analysis results subsequently fed back to the user are the target analysis results with high accuracy, and accurate video detection results can be fed back to the user.

[0069] The confidence threshold is not limited and can be set according to actual use needs. For example, the confidence threshold is 0.6, 0.7, 0.8, 0.85, 0.9, etc.

[0070] For example, the reference analysis result of the target video frame includes the object information of each target object in the target video frame, the object information includes the target category of the target object and the confidence that the target object is the target category, and the target filtering strategy includes filtering the object information of the target object whose confidence is lower than the confidence threshold (the confidence threshold is 0.8): the target video frame includes the living room area and the balcony area, there are target objects A and target object B in the living room area, and there is target object C in the balcony area; the category of target object A is cat and the confidence that target object A is a cat is 0.85, the target object B is dog and the confidence that target object B is a dog is 0.9, the target object C is bird and the confidence that target object C is a bird is 0.6; because the confidence that target object C is a bird is lower than the confidence threshold, the reference analysis result of target object C is filtered out; then, the initial analysis result is the relevant analysis result about target object A and target object B.

[0071] See also Figure 5 , Figure 5 yes Figure 1 It should be noted that if there is substantially the same result, this embodiment does not use Figure 5 The process sequence shown is limited. Figure 5 As shown, the initial analysis result of the target video frame includes the object information of each target object in the target video frame, the object information of the target object includes the target category of the target object and the position of the target object, the historical analysis result of the historical video frame includes the object information of each historical object in the historical video frame, the object information of the historical object includes the historical category of the historical object and the position of the historical object, and this embodiment includes:

[0072] Step S51: for each target object in the target video frame, determine the position coincidence degree between the target object and the historical objects of the same category.

[0073] In this embodiment, for each target object in the target video frame, the position coincidence degree of the target object and the historical object of the same category is determined. The position of the target object in the target video frame and the position of the historical object of the same category in the historical video frame are used to determine the corresponding coincidence degree of the target object; if the corresponding coincidence degree of the target object is high, it indicates that the position of the target object in the target video frame and the position of the historical object of the same category in the historical video frame are basically coincident, further indicating that the target object has not moved and is a static target.

[0074] For example, taking the target object as a vehicle: determine the position of the vehicle in the target video frame and determine the position of the vehicle in the historical video frame; use the position of the vehicle in the target video frame and the position of the vehicle in the historical video frame to determine the corresponding position coincidence of the vehicle; if the corresponding position coincidence of the vehicle is high, it indicates that the position of the vehicle in the target video frame and the position in the historical video frame are basically coincident, further indicating that the vehicle has not moved and the vehicle is a static target.

[0075] In one embodiment, the representation value of the position of the object includes the smallest outer rectangular box of the object. Figure 6 As shown, Figure 6 is a schematic diagram of an embodiment of the minimum outer rectangular frame of an object provided by the present application, and the historical analysis results of the historical video frames ( Figure 6 Detection results of video frame 1 in the historical object ( Figure 6 The position of the detection target 1) in is the smallest outer rectangular box of the historical object ( Figure 6 The small rectangular box on the left in the figure); the initial analysis result of the target video frame ( Figure 6 Detection results of the video frame 2 in the target object ( Figure 6 The position of the detection target 2) in the image is the smallest outer rectangular frame of the target object ( Figure 6 The small rectangular box on the right of the image).

[0076] Furthermore, if Figure 7 As shown, Figure 7 is a schematic diagram of an embodiment of the position overlap provided by the present application, using the minimum outer rectangular frame of the historical object ( Figure 7 The rectangular box corresponding to the detection target 1 in the image) and the minimum bounding rectangular box of the target object ( Figure 7 The rectangular box corresponding to the detection target 2 is determined, and the overlap between the minimum circumscribed rectangular box of the historical object and the minimum circumscribed rectangular box of the target object is determined as the position overlap between the target object and the historical objects of the same category.

[0077] It should be noted that there can be only one historical video frame, and the position of the target object in this historical video frame and the position of the target object in the target video frame are used to determine the position coincidence of the target object. If there are two or more historical video frames, the position of the target object in each historical video frame and the position of the target object in the target video frame are used to determine the position coincidence of the target object in each historical video frame.

[0078] Step S52: Determine whether the target object is a static target based on the position coincidence corresponding to the target object.

[0079] In this embodiment, whether the target object is a static target is determined based on the position coincidence corresponding to the target object. If the coincidence corresponding to the target object is high, it indicates that the position of the target object in the target video frame and the position of the historical object of the same category in the historical video frame are basically coincident, further indicating that the target object has not moved and is a static target.

[0080] In one embodiment, based on the position coincidence corresponding to the target object, it is determined whether the target object is a static target, specifically: in response to the position coincidence corresponding to the target object being greater than or equal to the coincidence threshold, the target object is determined to be a static target; in response to the position coincidence of the target object being less than the coincidence threshold, the target object is determined to be a dynamic target. In other words, a coincidence threshold is set, that is, a critical coincidence for determining a dynamic target or a static target is set; when the coincidence of the target object is greater than or equal to the coincidence threshold, it is considered that the coincidence corresponding to the target object is high, indicating that the position of the target object in the target video frame and the position of the same category of historical objects in the historical video frame are basically coincident, further indicating that the target object has not moved and the target object is a static target; when the coincidence of the target object is less than the coincidence threshold, it is considered that the coincidence corresponding to the target object is low, indicating that the position of the target object in the target video frame and the position of the same category of historical objects in the historical video frame do not coincide, further indicating that the target object moves and the target object is a dynamic target.

[0081] There is no limitation on the size of the overlap threshold, and it can be set according to actual use needs.

[0082] Step S53: Delete the object information corresponding to each static target from the initial analysis result to obtain a reference analysis result.

[0083] In this embodiment, the object information corresponding to each static target is deleted from the initial analysis result to obtain a reference analysis result. In other words, the analysis results related to the static target are deleted from the initial analysis result, and only the analysis results with dynamic targets are fed back to the user later, so that the user can intuitively understand the occurrence of the event.

[0084] See also Figure 8 , Figure 8It is a structural diagram of an embodiment of a video detection device provided by the present application. The video detection device 80 includes an analysis module 81, an acquisition module 82 and a filtering module 83. The analysis module 81 is used for the target node device to analyze the target video frame to obtain the initial analysis result of the target video frame, wherein the target video frame is reported to the target edge node device by the video acquisition device; and the acquisition module 82 is used to obtain the historical analysis result of at least one historical video frame; wherein the target video frame and at least one historical video frame are derived from the same target video, and the analysis results of each video frame of the target video are obtained by analyzing and executing at least two edge node devices, and the storage location of the analysis results of each video frame of the target video is the same; the filtering module 83 is used to use the historical analysis results of at least one historical video frame to perform static target filtering on the initial analysis result to obtain the target analysis result.

[0085] Among them, the initial analysis results of the target video frame include object information of each target object in the target video frame, and the object information of the target object includes the target category of the target object and the position of the target object. The historical analysis results of the historical video frame include object information of each historical object in the historical video frame, and the object information of the historical object includes the historical category of the historical object and the position of the historical object; the filtering module 83 is used to use the historical analysis results of at least one historical video frame to perform static target filtering on the initial analysis results to obtain target analysis results, including: for each target object in the target video frame, determining the position overlap between the target object and the historical object of the same category; based on the corresponding position overlap of the target object, determining whether the target object is a static target; from the initial analysis results, deleting the object information corresponding to each static target to obtain a reference analysis result.

[0086] Among them, the characterization value of the position of the above-mentioned object includes the minimum circumscribed rectangular frame of the object; and / or, the filtering module 83 is used to determine whether the target object is a static target based on the position overlap corresponding to the target object, including: in response to the position overlap corresponding to the target object being greater than or equal to the overlap threshold, determining that the target object is a static target; in response to the position overlap of the target object being less than the overlap threshold, determining that the target object is a dynamic target.

[0087] The at least one historical video frame is a video frame adjacent to the target video frame in the target video.

[0088] Among them, each edge node device is used to process at least two video frames in the target video, wherein the target video frame includes the first video frame processed by the target edge node device, and the corresponding historical video frame is a video frame processed by any other edge node device; and / or, the target video frame includes a non-first video frame processed by the target edge node device, and the corresponding historical video frame is a video frame processed by the target edge node device.

[0089] The video detection device 80 further includes a feedback module, and the feedback module is used to feedback the target analysis result to the user in response to the target analysis result including the analysis result of the dynamic target.

[0090] The analysis module 81 is used to analyze the target video frame to obtain the initial analysis result of the target video frame, including: analyzing the target video frame to obtain the reference analysis result of the target video frame; using the target filtering strategy to filter the reference analysis result to obtain the initial analysis result.

[0091] Among them, the above-mentioned target filtering strategy includes filtering the reference analysis results in the target video frame that do not belong to the user's area of ​​interest; and / or, the reference analysis results of the target video frame include object information of each target object in the target video frame, the object information includes the target category of the target object and the confidence that the target object is the target category, and the target filtering strategy includes filtering the object information of the target object whose confidence is lower than the confidence threshold.

[0092] The video detection device 80 further includes a concentration module 84, which is used to obtain video frames with dynamic targets; condense video frames corresponding to dynamic targets of the same category into target videos of corresponding categories; and provide feedback to the user on target videos corresponding to dynamic targets of each category.

[0093] See also Fig. 9 , Fig. 9 : is a schematic diagram of the structure of an embodiment of an electronic device provided by the present application. The electronic device 90 includes a memory 91 and a processor 92 coupled to each other, and the processor 92 is used to execute program instructions stored in the memory 91 to implement the steps of any of the above-mentioned video detection method embodiments. In a specific implementation scenario, the electronic device 90 may include but is not limited to: a microcomputer, a server, and in addition, the electronic device 90 may also include a mobile device such as a laptop computer and a tablet computer, which is not limited here.

[0094] Specifically, the processor 92 is used to control itself and the memory 91 to implement the steps of any of the above-mentioned video detection method embodiments. The processor 92 can also be called a CPU (Central Processing Unit). The processor 92 may be an integrated circuit chip with signal processing capabilities. The processor 92 can also be a general-purpose processor, a digital signal processor (Digital Signal Processor, DSP), an application-specific integrated circuit (Application Specific Integrated Circuit, ASIC), a field-programmable gate array (Field-Programmable Gate Array, FPGA) or other programmable logic devices, discrete gates or transistor logic devices, discrete hardware components. The general-purpose processor can be a microprocessor or the processor can also be any conventional processor, etc. In addition, the processor 92 can be implemented by an integrated circuit chip.

[0095] See also Fig.10 , Fig.10 It is a structural diagram of an embodiment of a computer-readable storage medium provided by the present application. The computer-readable storage medium 100 of the embodiment of the present application stores a program instruction 101, and when the program instruction 101 is executed, it implements the method provided by any embodiment of the video detection method of the present application and any non-conflicting combination. Among them, the program instruction 101 can form a program file and be stored in the above-mentioned computer-readable storage medium 100 in the form of a software product, so that a computer device (which can be a personal computer, a server, or a network device, etc.) executes all or part of the steps of each implementation method of the present application. The aforementioned computer-readable storage medium 100 includes: various media that can store program codes, such as a USB flash drive, a mobile hard disk, a read-only memory (ROM, Read-Only Memory), a random access memory (RAM, Random Access Memory), a disk or an optical disk, or a terminal device such as a computer, a server, a mobile phone, and a tablet.

[0096] If the technical solution of this application involves personal information, the product using the technical solution of this application has clearly informed the personal information processing rules and obtained the individual's voluntary consent before processing the personal information. If the technical solution of this application involves sensitive personal information, the product using the technical solution of this application has obtained the individual's separate consent before processing the sensitive personal information, and at the same time meets the "explicit consent" requirement. For example, on personal information collection devices such as cameras, clear and prominent signs are set to inform that the personal information collection scope has been entered and personal information will be collected. If the individual voluntarily enters the collection scope, it is deemed that he or she agrees to the collection of his or her personal information; or on the device that processes personal information, the personal information processing rules are notified by obvious signs / information, and the individual's authorization is obtained through pop-up information or by asking the individual to upload his or her personal information; among them, the personal information processing rules may include information such as the personal information processor, the purpose of personal information processing, the processing method, and the type of personal information processed.

[0097] The above description is only an implementation method of the present application, and does not limit the patent scope of the present application. Any equivalent structure or equivalent process transformation made using the contents of the present application specification and drawings, or directly or indirectly used in other related technical fields, are also included in the patent protection scope of the present application.

Claims

1. A video detection method, characterized in that: The method comprises: The target edge node device analyzes the target video frame to obtain an initial analysis result of the target video frame; wherein the target video frame is reported to the target edge node device by the video acquisition device; And, obtaining a historical analysis result of at least one historical video frame; wherein the target video frame and the at least one historical video frame are derived from the same target video, the analysis results of each video frame of the target video are obtained by performing analysis and execution by at least two edge node devices, and the storage location of the analysis results of each video frame of the target video is the same; The historical analysis result of the at least one historical video frame is used to perform static target filtering on the initial analysis result to obtain a target analysis result.

2. The method according to claim 1, characterized in that The initial analysis result of the target video frame includes object information of each target object in the target video frame, the object information of the target object includes the target category of the target object and the position of the target object, the historical analysis result of the historical video frame includes object information of each historical object in the historical video frame, the object information of the historical object includes the historical category of the historical object and the position of the historical object; the using the historical analysis result of the at least one historical video frame to perform static target filtering on the initial analysis result to obtain the target analysis result includes: For each target object in the target video frame, determining a position overlap between the target object and a historical object of the same category; Determining whether the target object is a static target based on the position coincidence corresponding to the target object; The object information corresponding to each of the static targets is deleted from the initial analysis result to obtain the target analysis result.

3. The method according to claim 2, characterized in that The representation value of the position of the object includes the minimum circumscribed rectangular frame of the object; And / or, determining whether the target object is a static target based on the position coincidence corresponding to the target object includes: In response to the position coincidence corresponding to the target object being greater than or equal to a coincidence threshold, determining that the target object is the static target; In response to the position coincidence of the target object being less than the coincidence threshold, the target object is determined to be the dynamic target.

4. The method according to claim 1, characterized in that: The at least one historical video frame is a video frame adjacent to the target video frame in the target video.

5. The method according to claim 4, characterized in that Each of the edge node devices is used to process at least two video frames in the target video, wherein: The target video frame includes the first video frame processed by the target edge node device, and the corresponding historical video frame is a video frame processed by any other edge node device; and / or, the target video frame includes a non-first video frame processed by the target edge node device, and the corresponding historical video frame is a video frame processed by the target edge node device.

6. The method according to claim 1, characterized in that The method further comprises: In response to the target analysis result including the analysis result of the dynamic target, the target analysis result is fed back to the user.

7. The method according to claim 1, characterized in that The analyzing the target video frame to obtain an initial analysis result of the target video frame includes: Analyze the target video frame to obtain a reference analysis result of the target video frame; The reference analysis result is filtered using a target filtering strategy to obtain the initial analysis result.

8. The method according to claim 7, characterized in that The target filtering strategy includes filtering reference analysis results in the target video frame that do not belong to the user's area of ​​interest; and / or the reference analysis results of the target video frame include object information of each target object in the target video frame, the object information includes the target category of the target object and the confidence that the target object is of the target category, and the target filtering strategy includes filtering object information of target objects whose confidence is lower than a confidence threshold.

9. The method according to claim 1, characterized in that: The method further comprises: Get the video frame with dynamic objects; Condensing the video frames corresponding to the dynamic targets of the same category into target videos of the corresponding category; Feedback target videos corresponding to each category of dynamic targets to the user.

10. An electronic device, characterized in that: The electronic device comprises a memory and a processor, wherein the memory stores program instructions, and the program instructions can be executed to implement the video detection method according to any one of claims 1 to 9.

11. A computer-readable storage medium, characterized in that: The computer-readable storage medium stores program instructions, and the program instructions can be executed to implement the video detection method according to any one of claims 1 to 9.

Images