Video processing method, device and computer-readable storage medium

Abstract

This application discloses a video processing method, device, and computer-readable storage medium. The method comprises: obtaining a video to be processed, the video to be processed comprising a plurality of video frames sequentially arranged in time; sequentially evaluating the quality of an object in each video frame as the current frame to obtain a quality evaluation result; executing a first allocation process if the quality evaluation result of the current frame is qualified; and extracting a video frame sequence of the object from a second qualified queue. This method can extract a video frame sequence of the object of qualified quality from the video to be processed, while reducing computational complexity.

Description

Technical Field

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

[0002] Image / video frames are a crucial source of information for human visual perception and intelligent visual analysis (such as target attribute recognition and behavior analysis). Video frame quality plays a crucial role in the adequacy and accuracy of the information obtained. For example, poor license plate quality in a video frame results in significant loss of license plate character information, making it difficult to recognize the characters and resulting in low accuracy. Similarly, poor quality of people in a video frame makes it impossible to identify them through human vision.

[0003] Therefore, in order to meet the needs of human visual perception and intelligent visual analysis, it is necessary to extract video frame sequences of qualified targets from the video. Summary of the Invention

[0004] The present application provides a video processing method, device and computer-readable storage medium, which can extract a video frame sequence of a target of qualified quality from a video.

[0005] To solve the above technical problems, a technical solution adopted by the present application is to provide a video processing method. The method includes: obtaining a video to be processed, the video to be processed including a number of video frames arranged continuously in time sequence; taking each video frame as the current frame in turn, performing a quality assessment on the target in the current frame to obtain a quality assessment result; if the quality assessment result of the current frame is qualified, executing a first allocation process, the first allocation process including: determining whether the historical frames arranged continuously with the current frame are in a first qualified queue or a second qualified queue; if the historical frames arranged continuously with the current frame are in the first qualified queue and the number of historical frames arranged continuously with the current frame is greater than a first quantity threshold, transferring the current frame and the historical frames arranged continuously with the current frame to the second qualified queue; if the historical frames arranged continuously with the current frame are in the second qualified queue, adding the current frame to the second qualified queue; if the historical frames arranged continuously with the current frame are in the second qualified queue and the number of historical frames arranged continuously with the current frame is not greater than the first quantity threshold, adding the current frame to the first qualified queue; extracting the target video frame sequence from the second qualified queue.

[0006] To solve the above technical problems, another technical solution adopted in this application is: providing a video processing device, which includes a processor and a memory connected to the processor, wherein the memory stores program instructions; the processor is used to execute the program instructions stored in the memory to implement the above method.

[0007] In order to solve the above technical problems, another technical solution adopted in this application is: providing a computer-readable storage medium storing program instructions, which can implement the above method when executed.

[0008] Through the above-mentioned method, when the quality evaluation result of the current frame is in a qualified state, the present application selectively includes the current frame in the first qualified queue, or the second qualified queue, or transfers the current frame and the historical frame from the first qualified queue to the second qualified queue based on the queue (first qualified queue or second qualified queue) and the number of historical frames set continuously with the current frame, thereby realizing queue management of video frames with qualified quality evaluation results, and then directly extracting a video frame sequence of a target with qualified quality (quality evaluation result is qualified state) from the second qualified queue, without having to select a video frame sequence of a target with qualified quality from the video to be processed after the quality evaluation. Therefore, while the present application can realize the extraction of a video frame sequence of a target with qualified quality from the video to be processed, it can also reduce the amount of calculation. BRIEF DESCRIPTION OF THE DRAWINGS

[0009] Figure 1 This is a flow chart of an embodiment of the video processing method of the present application;

[0010] Figure 2 yes Figure 1 Specific process diagram of S14;

[0011] Figure 3 yes Figure 1 Specific process diagram of S16;

[0012] Figure 4 This is a flow chart of another embodiment of the video processing method of the present application;

[0013] Figure 5 It is a schematic diagram of the application scenario of this application;

[0014] Figure 6 It is a structural diagram of the video processing system;

[0015] Figure 7 This is a flowchart of a specific example of the video processing method of the present application;

[0016] Figure 8 This is a structural diagram of an embodiment of a video processing device of the present application;

[0017] Figure 9 It is a structural diagram of an embodiment of a computer-readable storage medium of the present application. DETAILED DESCRIPTION

[0018] The following will be combined with the drawings in the embodiments of this application to clearly and completely describe the technical solutions in the embodiments of this application. Obviously, the embodiments described are only part of the embodiments of this application, not all of the embodiments. Based on the embodiments in this application, all other embodiments obtained by ordinary technicians in this field without making creative efforts are within the scope of protection of this application.

[0019] The terms "first," "second," and "third" in this application are used for descriptive purposes only and should not be understood as indicating or implying relative importance or implicitly indicating the number of the technical features indicated. Therefore, a feature specified as "first," "second," or "third" may explicitly or implicitly include at least one of the features. In the description of this application, "plurality" means at least two, for example, two, three, etc., unless otherwise specifically specified.

[0020] Reference herein to an "embodiment" means that a particular feature, structure, or characteristic described in connection with the embodiment may be included in at least one embodiment of the present application. The appearance of this phrase in various places in the specification does not necessarily refer to the same embodiment, nor does it refer to independent or alternative embodiments that are mutually exclusive of other embodiments. It is understood, both explicitly and implicitly, by those skilled in the art that the embodiments described herein may be combined with other embodiments unless there is a conflict.

[0021] Figure 1 It is a flow chart of an embodiment of the video processing method of the present application. It should be noted that if there is substantially the same result, this embodiment does not use Figure 1 The process sequence shown is limited. Figure 1 As shown, this embodiment may include:

[0022] S11: Obtain the video to be processed.

[0023] The video to be processed includes a plurality of video frames that are sequentially arranged in time.

[0024] The execution subject of this embodiment is a video processing device, which can be the camera itself, a computer, a mobile phone, a processor, etc. that establishes a communication connection with the camera. The video to be processed is obtained by the camera capturing the application scene.

[0025] S12: Taking each video frame as the current frame in turn, perform quality assessment on the target in the current frame to obtain a quality assessment result.

[0026] Targets can be any object requiring quality assessment (either human visual perception or intelligent visual analysis), such as people, cars, animals (dogs, pigs), license plates, or clothing. Target tracking can be performed on the video being processed, obtaining the target's trajectory. This trajectory can include the target's ID, its position in each video frame, its category, and more.

[0027] The basis for quality assessment may include the completeness of the target, the position of the target, the exposure of the target, the blur of the target, etc. Accordingly, the quality assessment may include the completeness assessment of the target, the position assessment of the target, the exposure assessment of the target, the blur assessment of the target, etc. Accordingly, the quality assessment results may include the completeness assessment result (target complete, target incomplete), the position assessment result (target position meets the preset position requirements, target position does not meet the preset position requirements), the exposure assessment result (target exposure is abnormal, target exposure is normal), and the blur assessment result (target is blurred, target is not blurred).

[0028] The complete evaluation of the target may include at least one of a complete evaluation based on a neural network and a complete evaluation based on an intersection-and-union ratio. The complete evaluation based on a neural network is to classify the target using a neural network to obtain a complete evaluation result based on a neural network. The complete situation of the target can also be called the occlusion situation of the target. The complete evaluation based on the intersection-and-union ratio is to obtain the intersection-and-union ratio between the position of the target and the position of the occluder, and to obtain a complete evaluation result based on the intersection-and-union ratio based on the relationship between the intersection-and-union ratio and the intersection-and-union ratio threshold. For example, the area of the target position is A, the area of the occluder position is B, and the intersection-and-union ratio is It can be understood that the dual complete evaluation using a neural network and intersection-over-union can improve the accuracy of the complete evaluation results.

[0029] The target's position meets the preset position requirements, meaning it's not within a specific region. For example, the specific region is the boundary of the current frame. If the target is within the boundary, it means it has just appeared in the field of view of the current frame or is about to disappear. The target's position can be evaluated through neural network classification or by calculating the distance between the target's position and the boundary. The specific implementation is similar to the complete target evaluation and is not detailed here.

[0030] The fuzzy evaluation and exposure evaluation of the target can be achieved through neural networks and other methods.

[0031] S13: Determine whether the quality evaluation result of the current frame is qualified or unqualified.

[0032] Constraints for a qualified quality assessment result for the current frame include the following: target integrity, target position meeting preset requirements, target exposure being correct, and target blur. For example, quality assessments include neural network-based integrity assessments and intersection-of-union (IoU)-based integrity assessments. If both the neural network-based and IoU-based integrity assessments indicate target integrity, the quality assessment result is qualified; otherwise, the quality assessment result is unqualified.

[0033] If the quality evaluation result of the current frame is in a qualified state, S14 to S15 are executed; if the quality evaluation result of the current frame is in an unqualified state, S16 is executed.

[0034] S14: Execute the first allocation process.

[0035] See also Figure 2 , S14 may include the following sub-steps:

[0036] S141: Determine whether the historical frame that is consecutive to the current frame is in the first qualified queue or the second qualified queue.

[0037] A historical frame refers to a video frame that precedes the current frame in the time sequence (chronological order) in the video to be processed, and a historical frame that is set continuously with the current frame refers to a video frame that is set continuously with the current frame in the video to be processed and precedes the current frame in time sequence. For example, the video to be processed includes video frames 1 to 10 that are set continuously in time sequence, the current frame is video frame 5, video frames 1 to 4 are historical frames, and video frame 4, or video frames 3 to 4, or video frames 2 to 4, or video frames 1 to 4 can all be called historical frames that are set continuously with the current frame. If video frame 4 is in the first qualified queue and video frame 3 is not in the first qualified queue, then the number of historical frames that are set continuously with the current frame is said to be in the first qualified queue, and the number of historical frames that are set continuously with the current frame is 1. If video frames 3 to 4 are in the first qualified queue, then the number of historical frames that are set continuously with the current frame is said to be in the first qualified queue, and the number of historical frames that are set continuously with the current frame is 2. If video frames 2 to 4 are in the first qualified queue, and video frame 1 is not in the first qualified queue, then the number of historical frames consecutive to the current frame in the first qualified queue is 3, and so on. The same applies to the second qualified queue.

[0038] If the historical frame set continuously with the current frame is in the first qualified queue, S142 to S144 are executed; if the historical frame set continuously with the current frame is in the second qualified queue, S145 is executed.

[0039] S142: Determine whether the number of historical frames set consecutively with the current frame is greater than a first number threshold.

[0040] If the number of packets is greater than the first threshold, execute S143; if the number of packets is not greater than the first threshold, execute S144.

[0041] S143: Transfer the current frame and the historical frames that are consecutive to the current frame to a second qualified queue.

[0042] For example, three historical frames consecutive to the current frame are in the first qualified queue, and the first quantity threshold is 2, then the current frame and the three historical frames consecutive to the current frame are transferred to the second qualified queue.

[0043] S144: Add the current frame to the first qualified queue.

[0044] S145: Add the current frame to the second qualified queue.

[0045] The video frames of the targets in the second qualified queue can be applied to human visual perception or intelligent visual analysis. It is understandable that in S141 to S145, when the quality assessment result of the current frame is in a qualified state and the historical frames (the number of which is greater than the first number threshold) set continuously with the current frame are in the first qualified queue, the current frame and the historical frames set continuously with the current frame are transferred to the second qualified queue. Thus, the process of transferring the quality assessment results in the first qualified queue to the second qualified queue is real-time, which facilitates the subsequent extraction of the target video frame sequence from the second qualified queue at any time, and breaks away from the limitation that the extraction can only be carried out after the quality assessment of all video frames of the video to be processed is complete.

[0046] S15: Extract the target video frame sequence from the second qualified queue.

[0047] Video frames that are consecutively arranged in the second qualified queue and whose number is greater than a third number threshold may be used as a target video frame sequence, where the third number threshold is greater than or equal to the first number threshold.

[0048] It can be understood that the third quantity threshold is greater than or equal to the first quantity threshold, which means that the timing of extracting the target video frame sequence from the second qualified queue is optional. In this way, the target video frame sequence can be extracted after the quality evaluation of all video frames of the video to be processed is completed, or after the quality evaluation of part of the video frames of the video to be processed is completed.

[0049] S16: Execute the second allocation process.

[0050] In some embodiments, the second allocation process may include: directly placing the current frame into a non-qualified queue.

[0051] In some embodiments, in conjunction with Figure 3 , S16 may include the following sub-steps:

[0052] S161: Determine whether the historical frame set continuously with the current frame is in the second qualified queue.

[0053] If the historical frame set continuously with the current frame is not in the second qualified queue, then execute S164. If the historical frame set continuously with the current frame is in the second qualified queue, then execute S162 to S164.

[0054] S162: Determine whether the number of historical frames that are consecutively set with the current frame and whose quality assessment results are unqualified is greater than a second number threshold.

[0055] If it is not greater than the second quantity threshold, execute S163; if it is greater than the second quantity threshold, execute S164.

[0056] S163: Add the current frame to the second qualified queue.

[0057] S164: The current frame is put into the unqualified queue.

[0058] It can be understood that, compared to the method of directly placing the current frame into the unqualified queue when the quality assessment result of the current frame is in a qualified state, in S161 to S164, the current frame is placed into the second qualified queue when the number of historical frames that are set continuously with the current frame and whose quality assessment results are unqualified is less than the second quantity threshold, otherwise the current frame is placed into the unqualified queue. The number of historical frames that are set continuously with the current frame and whose quality assessment results are unqualified is less than the second quantity threshold means that the error of the second qualified queue is within the allowable range. Therefore, through S161 to S164, the continuity of the second qualified queue in terms of timing can be improved while being within the allowable range of error.

[0059] Through the implementation of this embodiment, when the quality assessment result of the current frame is in a qualified state, the present application selectively includes the current frame in the first qualified queue, or includes it in the second qualified queue, or transfers the current frame and the historical frame from the first qualified queue to the second qualified queue based on the queue (first qualified queue or second qualified queue) and the number of historical frames set continuously with the current frame, thereby achieving queue management of video frames with qualified quality assessment results, and then directly extracting a video frame sequence of a target with qualified quality (quality assessment result is qualified state) from the second qualified queue, without having to select a video frame sequence of a target with qualified quality from the video to be processed after the quality assessment. Therefore, while the present application can achieve the extraction of a video frame sequence of a target with qualified quality from the video to be processed, it can also reduce the amount of calculation.

[0060] Figure 4 It is a flow chart of another embodiment of the video processing method of the present application. It should be noted that if there is substantially the same result, this embodiment does not Figure 4 The process sequence shown is limited. This embodiment is the steps that can be included before S162. Figure 4 As shown, this embodiment may include:

[0061] S21: Determine whether the unqualified state of the current frame is the first unqualified state or the second unqualified state.

[0062] The unqualified state may include a first unqualified state and a second unqualified state. Specifically, the constraints of the qualified state may be divided into hard constraints and soft constraints. In the second unqualified state, the hard constraints are not satisfied, and in the first unqualified state, the hard constraints are satisfied but the soft constraints are not satisfied.

[0063] For example, quality assessment includes a complete evaluation based on a neural network and a complete evaluation based on an intersection-and-union ratio. The intersection-and-union ratio complete evaluation result indicates that the target is complete, which is a hard constraint, while the neural network complete evaluation result indicates that the target is complete, which is a soft constraint. In the first unqualified state, the neural network complete evaluation result indicates that the target is incomplete, and the intersection-and-union ratio complete evaluation result indicates that the target is complete. In the second unqualified state, the intersection-and-union ratio complete evaluation result indicates that the target is incomplete.

[0064] For example, the quality assessment includes a target position assessment. The target position assessment result meeting a preset position requirement is a hard constraint. The target position assessment result in the first unqualified state is that the target position meets the preset position requirement, while the target position assessment result in the second unqualified state is that the target position meets the preset position requirement.

[0065] For another example, the quality assessment includes a complete evaluation based on a neural network, a complete evaluation based on an intersection-and-union ratio, and a target position evaluation. The target position evaluation result meets the preset position requirements, and the complete evaluation result based on the neural network is that the target is complete as a hard constraint, and the complete evaluation based on the intersection-and-union ratio is a soft constraint. In the second unqualified state, the target position evaluation result does not meet the preset position requirements, and the complete evaluation result based on the neural network is that the target is incomplete. In the first unqualified state, the target position evaluation result meets the preset position requirements, and the complete evaluation result based on the neural network is that the target is complete, but the complete evaluation result based on the intersection-and-union ratio is that the target is incomplete.

[0066] If the unqualified state of the current frame is the first unqualified state, S22 is executed; if the unqualified state of the current frame is the second unqualified state, S23 is executed.

[0067] S22: Execute S162 to S164.

[0068] S23: The current frame is put into the unqualified queue.

[0069] Combine as follows Figures 5-7 , the video processing method provided by this application is described in detail in the form of an example:

[0070] Figure 5 This is a schematic diagram of the application scenario of this application, such as Figure 5 As shown in the figure, the application scenario includes Person 1, Person 2, and a tree. Person 2 is the target, and the tree is the obstruction. Person 2 wants to walk from the left side of the tree to the right side of the tree. When Person 2 passes the tree, he will be obstructed by the tree.

[0071] Figure 6 This is a schematic diagram of the structure of the video processing system of this application. Figure 6 As shown, the video processing system includes an imaging device 11 and a video processing device 12. The imaging device 11 is used to capture the application scene to obtain a video to be processed, and the video processing device 12 is used to obtain the video to be processed (including 100 consecutive video frames in time sequence) from the imaging device 11 and process the video to be processed to obtain a video processing result.

[0072] See also Figure 7 , the processing process of the video processing device 12 may include:

[0073] 1) Perform target tracking on the video to be processed and obtain the trajectory of person 1 (person 1's ID and the position of person 1 in each video frame).

[0074] 2) Each video frame is sequentially used as the current frame, and quality assessment is performed on the objects in the current frame (including a complete evaluation based on the neural network and a complete evaluation based on the intersection-over-union ratio), obtaining a quality assessment result for the current frame. Hereinafter, the complete evaluation result based on the neural network is referred to as the first complete evaluation result, and the complete evaluation result based on the intersection-over-union ratio is referred to as the second complete evaluation result.

[0075] 3) Determine whether the quality assessment result of the current frame is qualified (the first complete assessment result and the second complete assessment result are both target complete) or unqualified (the first complete assessment result and / or the second complete assessment result are target incomplete). If qualified, proceed to 4); if unqualified, proceed to 8);

[0076] 4) Determine whether the historical frames consecutive to the current frame are in the first qualified queue, the second qualified queue, or the unqualified queue. If they are in the first qualified queue and the number of historical frames is greater than 3, proceed to 8)-a and 8)-b; if they are in the first qualified queue and the number of historical frames consecutive to the current frame is not greater than 3, proceed to 9); if they are in the unqualified queue, proceed to 10).

[0077] 5) Determine whether the historical frame set continuously with the current frame is in the second qualified queue. If not, proceed to 10); if in the second qualified queue, proceed to 6).

[0078] 6) Determine whether the current frame's unqualified status is the first unqualified status (the first complete evaluation result indicates an incomplete target, and the second complete evaluation result indicates a complete target) or the second unqualified status. If the first unqualified status is the second unqualified status (the second complete evaluation result indicates an incomplete target), proceed to 7); if the second unqualified status is the second unqualified status, proceed to 10).

[0079] 7) Determine whether the number of historical frames that are consecutive to the current frame and whose quality assessment results are unqualified is greater than 3. If greater than 3, proceed to 10); if not greater than 3, proceed to 8)-a.

[0080] 8)-a: Include the current frame into the second qualified sequence.

[0081] 8)-b: Transfer the historical frames that are set consecutively with the current frame to the second eligible queue.

[0082] 9) Add the current frame to the first eligible queue.

[0083] 10) The current frame is put into the unqualified queue.

[0084] 11) Extract the target video frame sequence from the second qualified queue.

[0085] Figure 8 This is a schematic diagram of the structure of an embodiment of the video processing device of the present application. Figure 8 As shown, the video processing device includes a processor 21 and a memory 22 coupled to the processor 21 .

[0086] The memory 22 stores program instructions for implementing the method of any of the above embodiments; the processor 21 is used to execute the program instructions stored in the memory 22 to implement the steps of the above method embodiments. The processor 21 can also be called a CPU (Central Processing Unit). The processor 21 may be an integrated circuit chip with signal processing capabilities. The processor 21 can also be a general-purpose processor, a digital signal processor (DSP), an application-specific integrated circuit (ASIC), a field programmable gate array (FPGA) or other programmable logic device, a discrete gate or transistor logic device, or a discrete hardware component. The general-purpose processor can be a microprocessor or the processor can also be any conventional processor, etc.

[0087] Figure 9 This is a schematic diagram of the structure of an embodiment of the computer-readable storage medium of the present application. Figure 9As shown, the computer-readable storage medium 30 of the embodiment of the present application stores program instructions 31, and when the program instructions 31 are executed, the method provided in the above embodiment of the present application is implemented. Among them, the program instructions 31 can form a program file and be stored in the above-mentioned computer-readable storage medium 30 in the form of a software product, so that a computer device (which can be a personal computer, server, or network device, etc.) or a processor (processor) executes all or part of the steps of the various embodiments of the present application. The aforementioned computer-readable storage medium 30 includes: various media that can store program codes, such as a USB flash drive, a mobile hard disk, a read-only memory (ROM), a random access memory (RAM), a magnetic disk or an optical disk, or terminal devices such as a computer, a server, a mobile phone, and a tablet.

[0088] In the several embodiments provided in this application, it should be understood that the disclosed systems, devices and methods can be implemented in other ways. For example, the device embodiments described above are merely schematic. For example, the division of units is only a logical function division. In actual implementation, there may be other division methods, such as multiple units or components can be combined or integrated into another system, or some features can be ignored or not executed. Another point is that the mutual coupling or direct coupling or communication connection shown or discussed can be an indirect coupling or communication connection through some interface, device or unit, which can be electrical, mechanical or other forms.

[0089] In addition, each functional unit in each embodiment of the present application can be integrated into one processing unit, or each unit can exist physically separately, or two or more units can be integrated into one unit. The above-mentioned integrated unit can be implemented in the form of hardware or in the form of a software functional unit. The above 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 content of the description and drawings of this application, 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 processing method, characterized in that: include: Acquire a video to be processed, where the video to be processed includes a plurality of video frames arranged consecutively in time sequence; Taking each of the video frames as a current frame in turn, performing a quality assessment on an object in the current frame to obtain a quality assessment result, wherein the quality assessment includes at least one of an integrity assessment of the object, a position assessment of the object, an exposure assessment of the object, and a blur assessment of the object; If the quality evaluation result of the current frame is in a qualified state, a first allocation process is executed, and the first allocation process includes: Determining whether a historical frame that is consecutive to the current frame is in a first qualified queue or a second qualified queue; If the historical frames that are consecutive to the current frame are in the first qualified queue, and the number of the historical frames that are consecutive to the current frame is greater than a first number threshold, transferring the current frame and the historical frames that are consecutive to the current frame to the second qualified queue; If the historical frame that is consecutive to the current frame is in the second qualified queue, then the current frame is included in the second qualified queue; If the historical frames that are consecutive to the current frame are in the first qualified queue, and the number of the historical frames that are consecutive to the current frame is not greater than the first number threshold, then adding the current frame to the first qualified queue; If the quality evaluation result of the current frame is unqualified, a second allocation process is executed, and the second allocation process includes: placing the current frame into a failed queue; or determining whether a historical frame that is consecutive to the current frame is in the second qualified queue; if the historical frame that is consecutive to the current frame is not in the second qualified queue, placing the current frame in the unqualified queue; The target video frame sequence is extracted from the second qualified queue.

2. The method according to claim 1, characterized in that The quality assessment result is that the target is complete or the target is incomplete. If the quality assessment result is that the target is complete, the quality assessment result is in a qualified state.

3. The method according to claim 2, characterized in that The quality assessment includes a complete assessment based on a neural network and a complete assessment based on an intersection-over-union ratio. If the complete assessment results based on a neural network and the complete assessment results based on an intersection-over-union ratio are both complete to the target, the quality assessment result is in a qualified state.

4. The method according to claim 1, wherein The second allocation process further includes: If the historical frames that are consecutively arranged with the current frame are in the second qualified queue, determining whether the number of historical frames that are consecutively arranged with the current frame and whose quality assessment results are unqualified is greater than a second number threshold; If the number is not greater than the second quantity threshold, the current frame is included in the second qualified queue; If the number is greater than the second quantity threshold, the current frame is included in the unqualified queue.

5. The method according to claim 4, characterized in that The unqualified state includes a first unqualified state and a second unqualified state; Before determining whether the number of historical frames that are consecutively set with the current frame and for which the quality assessment results are in the unqualified state reaches a second number threshold, the method further includes: Determining whether the unqualified state of the current frame is a first unqualified state or a second unqualified state; If the unqualified state of the current frame is the first unqualified state, performing the steps of determining whether the number of historical frames that are consecutively set with the current frame and whose quality assessment results are the unqualified state reaches a second number threshold and subsequent steps; If the unqualified state of the current frame is the second unqualified state, the current frame is included in the unqualified queue.

6. The method according to claim 5, characterized in that The quality assessment includes a complete assessment based on a neural network and a complete assessment based on an intersection-and-union ratio. In the first unqualified state, the complete assessment result based on the neural network is that the target is incomplete, and the complete assessment result based on the intersection-and-union ratio is that the target is complete. In the second unqualified state, the complete assessment result based on the intersection-and-union ratio is that the target is incomplete.

7. The method according to claim 5, characterized in that The quality assessment includes a position assessment of the target. In the first unqualified state, the position assessment result of the target is that the position of the target meets the preset position requirement. In the second unqualified state, the position assessment result of the target is that the position of the target does not meet the preset position requirement.

8. The method according to claim 1, characterized in that The extracting the target video frame sequence from the second qualified queue includes: The video frames in the second qualified queue that are continuously arranged and whose number is greater than a third number threshold are used as the target video frame sequence, wherein the third number threshold is greater than or equal to the first number threshold.

9. A video processing device, characterized in that: comprising a processor and a memory connected to the processor, wherein: The memory stores program instructions; The processor is configured to execute the program instructions stored in the memory to implement the method according to any one of claims 1 to 8.

10. A computer-readable storage medium, characterized in that The computer-readable storage medium stores program instructions, which can be executed by a processor and implement the method according to any one of claims 1 to 8 when executed.

Images