A video processing method, apparatus and terminal

Abstract

This application provides a video processing method, including the following steps: parsing the original video to extract each frame of the original video; selecting the current frame from each frame of the original video, and determining whether the current frame is the initial frame of the original video or an interval frame distributed in the original video at fixed time intervals; if the current frame is the initial frame or an interval frame, using a target detection model configured with a first image processing parameter group to find the sensitive region contained in the current frame, and using a target tracking device configured with a second image processing parameter group to erase the sensitive region contained in the current frame, wherein the image processing quality standard specified by the first image processing parameter group is higher than the image processing quality standard specified by the second image processing parameter group; if the current frame is not the initial frame or an interval frame, using a target tracking device configured with a second image processing parameter group to erase the sensitive region contained in the current frame. This solves the problem of incomplete removal of sensitive information in the prior art and improves the erasure efficiency.

Description

Technical Field

[0001] This application relates to the technical field of video image processing, specifically to a video processing method, apparatus, and terminal. Background Technology

[0002] With the rapid development of society, people are paying increasing attention to personal information security. my country has issued corresponding principles and regulations regarding video or image data recorded by in-vehicle devices. Specifically, there are two principles: the in-vehicle processing principle, which states that data should not be provided to entities outside the vehicle unless absolutely necessary; and the anonymization principle, which states that if it is necessary to provide data to entities outside the vehicle, anonymization and desensitization processing should be performed as much as possible. The regulations stipulate that any function that requires obtaining video or image data through in-vehicle devices and that involves sending the data to entities outside the vehicle or to the vehicle network will not be permitted without desensitization processing.

[0003] Against this backdrop, prior art (CN202110701474.9) proposes a method, apparatus, and electronic device for processing sensitive regions in images. This method uses a deep learning-based image recognition model to identify pixels containing sensitive information and then blurs or removes the information. However, prior art desensitizes each frame of the video at the pixel level, resulting in excessive computation, significant time consumption, and high hardware resource consumption, making it unsuitable for real-time video desensitization in in-vehicle devices.

[0004] Existing technology 2 (CN202210659749.1) proposes a method, apparatus, and computer-readable storage medium for removing sensitive information from videos. For mainstream h.264 format videos, it only decodes and desensitizes keyframes before re-encoding, without desensitizing P-frames and B-frames to improve desensitization efficiency. However, when sensitive information exists within P-frames or B-frames, existing technology 2 cannot achieve complete desensitization of the video. Summary of the Invention

[0005] In view of this, this application provides a video processing method, apparatus and terminal, which aims to solve the problem that sensitive information cannot be erased when it exists in P frames or B frames in the prior art, and also improves the efficiency of sensitive information erasure in video, so as to be suitable for the limited computing resources of vehicle equipment.

[0006] To achieve the above objectives, this application provides a video processing method, comprising the following steps:

[0007] The original video is parsed to extract each frame of the original video;

[0008] Select the current frame from each frame of the original video, and determine whether the current frame is the initial frame of the original video or an interval frame distributed in the original video at fixed time intervals.

[0009] If the current frame is an initial frame or an interval frame, a target detection model configured with a first image processing parameter group is used to find the sensitive region contained in the current frame, and a target tracking device configured with a second image processing parameter group is used to erase the sensitive region contained in the current frame, wherein the image processing quality standard specified by the first image processing parameter group is higher than the image processing quality standard specified by the second image processing parameter group.

[0010] If the current frame is not an initial frame or an interval frame, then a target tracking device configured with a second image processing parameter group is used to erase the sensitive region contained in the current frame.

[0011] Optionally, the step of using a target detection model configured with a first set of image processing parameters to find the sensitive regions contained in the current frame includes:

[0012] The target detection model is trained based on the image corresponding to the current frame to configure the target detection model to identify a preset type of tracking object and its location in the image corresponding to the current frame.

[0013] The tracking object and its location in the image corresponding to the current frame are selected using a first target bounding box used to mark the tracking object.

[0014] Optionally, the step of using a target detection model configured with a first set of image processing parameters to find the sensitive region contained in the current frame further includes:

[0015] The target detection model is controlled to determine the sensitive region contained in the image corresponding to the current frame based on the first target bounding box;

[0016] The sensitive region is selected using a second target bounding box used to mark the sensitive region, and the first target bounding box in the image corresponding to the current frame is erased.

[0017] Optionally, the step of using a target tracking device configured with a second set of image processing parameters to erase the sensitive region contained in the current frame includes:

[0018] The target tracking device is controlled to acquire the number and position of the second target bounding boxes in the image corresponding to the current frame, and the motion pattern and appearance features of the tracked object are acquired based on the number and position of the second target bounding boxes.

[0019] The target tracking device learns how to identify a preset type of tracking object and its location in the image corresponding to the current frame based on the motion pattern and appearance features, and after completing the learning, controls the target tracking device to erase the sensitive area selected by the second target box in the image corresponding to the current frame.

[0020] Optionally, the step of using a target tracking device configured with a second set of image processing parameters to erase the sensitive region contained in the current frame further includes:

[0021] The subsequent frames of the current frame are input into the target tracking device. The target tracking device determines the position of the tracked object in the image corresponding to the subsequent frame based on the current frame, and uses the second target bounding box to select the sensitive area contained in the tracked object based on the position of the tracked object.

[0022] Optionally, inputting subsequent frames of the current frame into the target tracking device includes:

[0023] If the subsequent frame is an interval frame, then determine whether there is at least one of a newly added second target box and a lost second target box in the image corresponding to the interval frame;

[0024] If so, the target detection model is controlled to perform at least one operation: correcting newly added second target boxes and deleting lost second target boxes.

[0025] Optionally, inputting subsequent frames of the current frame into the target tracking device further includes:

[0026] If the subsequent frame is an interval frame, and the preset type of tracking object is not identified in the image corresponding to the interval frame, then the target tracking device is used to identify the tracking object in the subsequent frames of the interval frame.

[0027] Optionally, the first image processing parameter set includes a first processing precision, a first processing speed, and a first processing time, and the second image processing parameter set includes a second processing precision, a second processing speed, and a second processing time, wherein the first processing precision, the first processing speed, and the first processing time are respectively greater than the second processing precision, the second processing speed, and the second processing time.

[0028] This application also provides a video processing apparatus, the apparatus comprising:

[0029] The video processing module is used to parse the original video to extract each frame of the original video;

[0030] A frame selection module, electrically connected to the video processing module, is used to select the current frame from each frame of the original video;

[0031] The frame count determination module is electrically connected to the current frame count selection module and is used to determine whether the current frame is the initial frame of the original video or an interval frame distributed in the original video at a fixed time interval.

[0032] The execution module, electrically connected to the current frame number determination module, is used to find the sensitive regions contained in the current frame using a target detection model configured with a first image processing parameter group when the current frame is an initial frame or an interval frame, and to erase the sensitive regions contained in the current frame using a target tracking device configured with a second image processing parameter group; or, when the current frame is not an initial frame or an interval frame, to erase the sensitive regions contained in the current frame using a target tracking device configured with a second image processing parameter group.

[0033] This application also provides a terminal, which includes a memory and a processor. The memory stores a computer program, and when the processor executes the computer program stored in the memory, it implements the video processing method described above.

[0034] This invention employs a hybrid video processing method to analyze the original video and extract each frame. When the current frame is neither the initial frame nor an interval frame, a target tracking device configured with a second set of image processing parameters is used to identify and erase the sensitive regions contained in the current frame. When the current frame is the initial frame or an interval frame, a target detection model configured with a first set of image processing parameters is used to identify the sensitive regions contained in the frame, and then the target tracking device is used to erase the sensitive regions contained in the current frame. This solves the problem in the prior art where sensitive information cannot be erased when it exists in P-frames or B-frames, and also improves the efficiency of sensitive information erasure in videos. It is suitable for situations where the computing resources of in-vehicle equipment are limited. Attached Figure Description

[0035] To more clearly illustrate the technical solutions of the embodiments of this application, the accompanying drawings used in the embodiments will be briefly introduced below. It should be understood that the following drawings only show some embodiments of this application and should not be regarded as a limitation of the scope. For those skilled in the art, other related drawings can be obtained based on these drawings without creative effort.

[0036] Figure 1 This is a schematic flowchart of a video processing method according to an embodiment of this application;

[0037] Figure 2 This is a schematic diagram of the structure of an apparatus for a video processing method according to an embodiment of this application;

[0038] Figure 3 This is a schematic diagram of the structure of a terminal for a video processing method according to an embodiment of this application. Detailed Implementation

[0039] The specific embodiments of the present invention will now be described in detail with reference to the accompanying drawings. Obviously, the described embodiments are merely some, not all, of the embodiments of the present invention. Based on the description of the present invention, all other embodiments obtained by those skilled in the art without inventive effort are within the scope of protection of the present invention.

[0040] In the description of this invention, the terms "first," "second," "third," etc., are used merely to distinguish elements with similar properties, and do not indicate or imply relative importance or a specific order.

[0041] Furthermore, the terms “including,” “comprising,” or any other variations thereof are intended to cover non-exclusive inclusion, which includes not only the elements listed but also other elements not expressly listed.

[0042] Figure 1 The following is a flowchart illustrating the video processing method provided in an embodiment of the present invention. Please refer to it. Figure 1 The video processing method includes the following steps:

[0043] S1. Analyze the original video to extract each frame. Specifically, the original video is the video that has not yet undergone sensitive content processing. The device used to acquire the original video can be a dashcam, smart surveillance camera, mobile phone, camera, tablet, etc., and is not limited to any one device. Analyze the original video to obtain all frames of the original video, so that sensitive information can be processed for each frame individually.

[0044] S2. Select the current frame from each frame of the original video.

[0045] S3. Determine whether the current frame is the initial frame of the original video or an interval frame distributed throughout the original video at fixed time intervals. Specifically, from all frames of the original video obtained in step S1, each frame is sequentially treated as the current frame and the corresponding method is executed. Then, it is determined whether the current frame is at least one of two special types of frames: the initial frame and the interval frame. The initial frame is the first frame among all frames of the original video, and the interval frame is a frame distributed throughout all frames of the original video at fixed time intervals. The fixed time interval can be one millisecond, one second, ten seconds, ten minutes, one hour, etc., and is not uniquely limited here.

[0046] S4. If the current frame is an initial frame or an interval frame, the target detection model configured with the first image processing parameter group is used to find the sensitive regions contained in the current frame, and the target tracking device configured with the second image processing parameter group is used to erase the sensitive regions contained in the current frame. The image processing quality standard specified by the first image processing parameter group is higher than that specified by the second image processing parameter group. If the current frame is not an initial frame or an interval frame, the target tracking device configured with the second image processing parameter group is used to erase the sensitive regions contained in the current frame. Specifically, when determining whether the current frame is an initial frame or an interval frame, the target detection model configured with the first image processing parameter group is used to find the sensitive regions contained in the current frame, and then the target tracking device configured with the second image processing parameter group is used to erase the sensitive regions contained in the current frame. Conversely, the target tracking device configured with the second image processing parameter group is used directly to erase the sensitive regions contained in the current frame. For example, the first and second image processing parameter groups include the same content, such as required energy consumption, recognition rate, calculation speed, calculation accuracy, and image clarity optimization.

[0047] In one embodiment, through steps S1 to S4 described above, each frame of the original video is parsed, and the current frame is determined to be either an initial frame or an interval frame. In either case, the sensitive regions contained in the current frame are removed using two different methods. Normally, a target tracking device configured with a second set of image processing parameters is directly used to remove the sensitive regions contained in the current frame. Although this sacrifices some computational accuracy, it significantly reduces the energy consumption and resources required for computation, maintaining low-power operation over a long period under normal conditions. However, in special cases (when the current frame is an initial frame or an interval frame), a target detection model with higher computational accuracy and greater energy consumption and resources is first invoked to find the sensitive regions contained in the current frame, and then the target tracking device is used to remove these sensitive regions. By repeating steps S1 to S4 to process the sensitive information of the current frame of the original video in sequence, the problem that sensitive information cannot be removed when it exists in a P-frame or B-frame in the prior art can be effectively solved. At the same time, through the cooperation of two video processing methods, this invention can improve the efficiency of removing sensitive information from the video, and is also applicable to vehicle computer systems with efficient hardware resources, achieving a balance between accuracy and computer resource usage.

[0048] Optionally, the step S4, which involves using a target detection model configured with a first set of image processing parameters to find the sensitive region contained in the current frame, specifically includes: training the target detection model based on the image corresponding to the current frame to configure it for identifying a preset type of tracking object and its location in the image corresponding to the current frame. Then, a first bounding box for marking the tracking object is used to select the tracking object and its location in the image corresponding to the current frame. Specifically, the target detection model is first trained using a real-time algorithm based on the content of the image corresponding to the current frame, enabling it to accurately and quickly identify the preset type of tracking object and its location in the image. For example, the preset type of tracking object can be animals, humans, cars, houses, signs, a large screen installed on the exterior wall of a shopping mall, etc., and is not limited to any single type. Real-time training enables the target detection model to use the first bounding box to select the identified tracking object and its location. For example, if the image corresponding to the current frame shows a cow in the middle of a road, the real-time trained target detection model can quickly and accurately identify it as a cow and use the first bounding box to select the cow and its location.

[0049] Optionally, the step S4, which involves using a target detection model configured with the first image processing parameter group to find the sensitive region contained in the current frame, further includes: controlling the target detection model to determine the sensitive region contained in the image corresponding to the current frame based on the first target box. The sensitive region is then selected using a second target box used to mark the sensitive region, and the first target box in the image corresponding to the current frame is removed. Specifically, the trained target detection model described above first determines whether the region selected by the first target box in the image corresponding to the current frame contains the sensitive region of the tracked object. If it does, the target detection model uses the second target box to select the sensitive region of the tracked object, and then only the second target box is retained to reduce the complexity of the image and improve the user experience. The sensitive region may include a face, important male or female body parts, a vehicle license plate, etc., and is not limited to a single region.

[0050] Optionally, the step S4, which involves using a target tracking device configured with a second image processing parameter group to erase sensitive regions contained in the current frame, specifically includes: controlling the target tracking device to acquire the number and position of second target boxes in the image corresponding to the current frame, and acquiring the motion patterns and appearance features of the tracked object based on the number and position of the second target boxes. The target tracking device learns how to identify a preset type of tracked object and its location in the image corresponding to the current frame based on the motion patterns and appearance features, and after completing the learning, controls the target tracking device to erase the sensitive regions selected by the second target boxes in the image corresponding to the current frame. Specifically, firstly, the target tracking device acquires the number and position of second target boxes selected by the target detection model in the sensitive regions of the image corresponding to the current frame; secondly, the target tracking device learns based on the number and position of the second target boxes to acquire the motion patterns and appearance features of the tracked object, thereby continuously improving and optimizing its ability to accurately and quickly identify a preset type of tracked object and its location in the image corresponding to the current frame; finally, the target tracking device erases the sensitive regions selected by the second target boxes in the image corresponding to the current frame. In this embodiment, the core of this step is to provide the results obtained in real time by the high-precision, high-energy-consumption target detection model to the target tracking device with lower-precision, low-power consumption for training and learning. This can continuously improve and optimize the computational accuracy of the target detection model, so that in special cases (when the current frame is the initial frame or an interval frame), the target detection model can give full play to its low-power advantage while ensuring recognition accuracy and efficiency, making it suitable for situations where the computing power resources of the vehicle-mounted device are limited.

[0051] Optionally, the step S4, which involves using a target tracking device configured with a second image processing parameter group to remove sensitive regions contained in the current frame, further includes: inputting subsequent frames of the current frame into the target tracking device; using the target tracking device to determine the position of the tracking object in the image corresponding to the subsequent frame based on the current frame; and using a second target bounding box to select the sensitive regions contained in the tracking object based on the position of the tracking object. Specifically, after training as described above, the target detection model is capable of desensitizing each frame of the original video under normal circumstances (when the current frame is not the initial frame or an interval frame). At this time, subsequent frames of the current frame are input into the target tracking device, and the target tracking device is used to determine whether a tracking object exists in the image corresponding to the subsequent frame. If it exists, the target tracking device can directly select the sensitive regions contained in the tracking object using the second target bounding box.

[0052] Optionally, inputting subsequent frames of the current frame into the target tracking device as described above specifically includes: if the subsequent frame is an interval frame, determining whether there is at least one of a newly added second target box and a lost second target box in the image corresponding to the interval frame; if so, controlling the target detection model to perform at least one operation of correcting the newly added second target box and deleting the lost second target box. Specifically, during the continuous repetition of steps S1 to S4, when it is identified that the subsequent frame of the current frame is an interval frame, the target detection model is used to obtain the image corresponding to the interval frame and determine whether there is a lost second target box that has not been selected within the sensitive area, or a newly added second target box that did not appear in the previous interval frame. If so, then the target detection model is controlled to delete the lost second target box or correct the newly added second target box. In this embodiment, the core of this step is to perform timely corrections and adjustments during the continuous repetition of steps S1 to S3. When the subsequent frame is an interval frame, in addition to performing the operations described above, the target detection model also needs to use its advantages of high computational accuracy and high recognition rate to correct or adjust the deviation of the second target box selected by the target tracking device in the image corresponding to the previous frame of the interval frame, so that the target tracking device can ensure that the second target box accurately selects the sensitive area contained in the tracked object in the next frame of the interval frame.

[0053] Optionally, the input of subsequent frames of the current frame to the target tracking device described above further includes: if the subsequent frame is an interval frame, and no preset type of tracking object is identified in the image corresponding to the interval frame, then the target tracking device is used to identify the tracking object in the subsequent frame of the interval frame. Specifically, during the continuous repetition of steps S1 to S4, when it is identified that the subsequent frame of the current frame is an interval frame, and no preset type of tracking object is identified in the image corresponding to the interval frame, the target tracking device is directly activated to identify the tracking object in the subsequent frame of the interval frame. This method can further reduce energy consumption.

[0054] Optionally, the first image processing parameter set includes a first processing precision, a first computing speed, and a first computing time, and the second image processing parameter set includes a second processing precision, a second computing speed, and a second computing time. The first processing precision, the first computing speed, and the first computing time are respectively greater than the second processing precision, the second computing speed, and the second computing time. In this embodiment, both the first and second image processing parameter sets include computational precision, computational speed, and computing time, and each set of parameters in the first image processing parameter set is greater than each set of parameters in the second image processing parameter set. That is, the computational precision, computational speed, and computing time of the target monitoring model are all superior to those of the target tracking device. Through the cooperation between the two, it is possible to improve the efficiency of erasing sensitive information in the video while also being suitable for an in-vehicle computer system that effectively utilizes hardware resources, achieving a balance between accuracy and computer resource consumption.

[0055] Figure 2 This diagram illustrates the structure of an apparatus for a video processing method according to an embodiment of the present invention. Those skilled in the art will understand that... Figure 2 This is merely an example of a device and does not constitute a limitation on the device.

[0056] Please refer to Figure 2 The device includes:

[0057] The video processing module 10 is used to parse the original video to extract each frame of the original video.

[0058] The frame selection module 20 is electrically connected to the video processing module 10 and is used to select the current frame from each frame of the original video.

[0059] The frame number determination module 30 is electrically connected to the current frame number selection module 20 and is used to determine whether the current frame is the initial frame of the original video or an interval frame distributed in the original video at fixed time intervals.

[0060] The execution module 40 is electrically connected to the current frame number determination module 30. When the current frame is an initial frame or an interval frame, it uses a target detection model configured with a first image processing parameter group to find the sensitive region contained in the current frame and uses a target tracking device configured with a second image processing parameter group to erase the sensitive region contained in the current frame. Alternatively, when the current frame is not an initial frame or an interval frame, it uses a target tracking device configured with a second image processing parameter group to erase the sensitive region contained in the current frame.

[0061] The aforementioned device and its internal modules work together to achieve the same beneficial effects as the video processing method described above.

[0062] Figure 3This diagram illustrates the structure of a terminal for a video processing method according to an embodiment of the present invention. The terminal can be a computing device such as a computer, laptop, handheld computer, or cloud server. The terminal may include, but is not limited to, a processor and a memory. Those skilled in the art will understand that... Figure 3 This is merely an example of a terminal and does not constitute a limitation on the terminal. It may include more or fewer components than shown, or combine certain components, or different components. For example, a terminal may also include input / output devices, network access devices, buses, etc.

[0063] Please refer to Figure 3 This application also provides a terminal, which includes a memory and a processor. The memory stores a computer program. When the processor executes the computer program stored in the memory, it implements the video processing method described above and has the same beneficial effects.

[0064] The above description is merely a specific embodiment of this application, but the scope of protection of this application is not limited thereto. Any variations or substitutions that can be easily conceived by those skilled in the art within the technical scope disclosed in this invention should be included within the scope of protection of this invention. Therefore, the scope of protection of this invention should be determined by the appended claims.

Claims

1. A video processing method, characterized in that, Includes the following steps: The original video is parsed to extract each frame of the original video; Select the current frame from each frame of the original video, and determine whether the current frame is the initial frame of the original video or an interval frame distributed in the original video at fixed time intervals. If the current frame is an initial frame or an interval frame, a target detection model configured with a first image processing parameter group is used to find the sensitive region contained in the current frame, and a target tracking device configured with a second image processing parameter group is used to erase the sensitive region contained in the current frame, wherein the image processing quality standard specified by the first image processing parameter group is higher than the image processing quality standard specified by the second image processing parameter group. If the current frame is not the initial frame or an interval frame, then the target tracking device configured with the second image processing parameter group is used to erase the sensitive area contained in the current frame; The step of using a target detection model configured with a first set of image processing parameters to find the sensitive regions contained in the current frame includes: The target detection model is trained based on the image corresponding to the current frame to configure the target detection model to identify a preset type of tracking object and its location in the image corresponding to the current frame. The target detection model uses a first target bounding box to select the tracked object and its location in the image corresponding to the current frame, so as to determine the sensitive region contained in the tracked object in the image corresponding to the current frame; The sensitive region is selected using a second target bounding box used to mark the sensitive region, and the first target bounding box in the image corresponding to the current frame is erased; The step of using a target tracking device configured with a second set of image processing parameters to erase the sensitive region contained in the current frame includes: The target tracking device is controlled to acquire the number and position of the second target bounding boxes in the image corresponding to the current frame, and the motion pattern and appearance features of the tracked object are acquired based on the number and position of the second target bounding boxes. The target tracking device learns how to identify a preset type of tracking object and its location in the image corresponding to the current frame based on the motion pattern and appearance features, and after completing the learning, controls the target tracking device to erase the sensitive area selected by the second target box in the image corresponding to the current frame; The subsequent frames of the current frame are input into the target tracking device. The target tracking device determines the position of the tracked object in the image corresponding to the subsequent frame based on the current frame, and uses the second target bounding box to select the sensitive area contained in the tracked object based on the position of the tracked object.

2. The method according to claim 1, characterized in that, The step of inputting subsequent frames of the current frame into the target tracking device includes: If the subsequent frame is an interval frame, then determine whether there is at least one of a newly added second target box and a lost second target box in the image corresponding to the interval frame; If so, the target detection model is controlled to perform at least one operation: correcting newly added second target boxes and deleting lost second target boxes.

3. The method according to claim 1, characterized in that, The step of inputting subsequent frames of the current frame into the target tracking device further includes: If the subsequent frame is an interval frame, and the preset type of tracking object is not identified in the image corresponding to the interval frame, then the target tracking device is used to identify the tracking object in the subsequent frames of the interval frame.

4. The method according to claim 1, characterized in that, The first image processing parameter set includes a first processing precision, a first processing speed, and a first processing time. The second image processing parameter set includes a second processing precision, a second processing speed, and a second processing time. The first processing precision, the first processing speed, and the first processing time are respectively greater than the second processing precision, the second processing speed, and the second processing time.

5. A video processing apparatus, characterized in that, The device includes: The video processing module is used to parse the original video to extract each frame of the original video; A frame selection module, electrically connected to the video processing module, is used to select the current frame from each frame of the original video; The frame count determination module is electrically connected to the current frame count selection module and is used to determine whether the current frame is the initial frame of the original video or an interval frame distributed in the original video at a fixed time interval. An execution module, electrically connected to the current frame number determination module, is used to find the sensitive region contained in the current frame using a target detection model configured with a first image processing parameter group when the current frame is an initial frame or an interval frame, and to erase the sensitive region contained in the current frame using a target tracking device configured with a second image processing parameter group; or, when the current frame is not an initial frame or an interval frame, to erase the sensitive region contained in the current frame using a target tracking device configured with a second image processing parameter group. The step of using a target detection model configured with a first set of image processing parameters to find the sensitive region contained in the current frame includes: training the target detection model based on the image corresponding to the current frame to configure the target detection model for identifying a preset type of tracking object and its location in the image corresponding to the current frame; the target detection model using a first target bounding box to mark the tracking object to select the tracking object and its location in the image corresponding to the current frame to determine the sensitive region contained in the tracking object in the image corresponding to the current frame; using a second target bounding box to mark the sensitive region to select the sensitive region and removing the first target bounding box in the image corresponding to the current frame; The step of using a target tracking device configured with a second image processing parameter group to erase the sensitive region contained in the current frame includes: controlling the target tracking device to obtain the number and position of the second target boxes in the image corresponding to the current frame, obtaining the motion pattern and appearance features of the tracked object based on the number and position of the second target boxes; using the target tracking device to learn how to identify a preset type of tracked object and its location in the image corresponding to the current frame based on the motion pattern and appearance features, and controlling the target tracking device to erase the sensitive region selected by the second target boxes in the image corresponding to the current frame after completing the learning; inputting subsequent frames of the current frame into the target tracking device, using the target tracking device to determine the position of the tracked object in the image corresponding to the subsequent frame based on the current frame, and using the second target boxes to select the sensitive region contained in the tracked object based on the position of the tracked object.

6. A terminal, characterized in that, The terminal includes a memory and a processor. The memory stores a computer program, and when the processor executes the computer program stored in the memory, it implements the video processing method as described in any one of claims 1 to 4.

Images