To-be-detected image screening method, device and equipment and computer readable storage medium

Abstract

The invention discloses a to-be-detected image screening method, device and equipment and a computer readable storage medium. The method comprises the following steps: determining an effective detection time period corresponding to each camera; acquiring video data acquired by each camera in the corresponding effective detection time period; for each path of video data, detecting whether a differential operation result of adjacent image frames in the video data is greater than a first threshold; and if the image frame is greater than the first threshold value, taking the image frame acquired later in the adjacent image frames as an image to be detected. According to the method and the device, the video data acquired by each camera in the corresponding effective detection time period is taken as a basis, and the frame difference algorithm is combined to screen the to-be-detected image so as to be used for a subsequent detection task, so that the data volume faced by the detection task is reduced, and the computing resources are saved.

Description

Technical Field

[0001] The present application relates to the field of artificial intelligence technology, and in particular to a method, device, equipment and computer-readable storage medium for screening images to be detected. Background Art

[0002] With the continuous development of artificial intelligence, the application of intelligent algorithms is also playing an increasingly important role in improving the efficiency of social governance. For example, by deploying cameras in key areas and combining intelligent algorithms, detection tasks can be carried out based on video data captured by cameras to discover possible risks.

[0003] However, when there are many video data paths to be detected, the amount of data faced by the detection task becomes very large, which requires high-performance hardware to provide sufficient computing resources, thereby increasing hardware costs. Summary of the invention

[0004] The present application provides a method, device, equipment and computer-readable storage medium for screening images to be detected, which can solve the technical problem of large data volume faced by detection tasks existing in the prior art.

[0005] In a first aspect, an embodiment of the present application provides a method for screening images to be detected, the method comprising: Determine the effective detection time period corresponding to each camera; Obtain the video data collected by each camera during its corresponding effective detection time period; For each channel of video data, detecting whether a difference operation result of adjacent image frames in the video data is greater than a first threshold; If it is greater than the first threshold, the image frame acquired later in the adjacent image frames is used as the image to be detected.

[0006] In combination with the first aspect, in one implementation, determining the effective detection time period corresponding to each camera includes: For each camera, construct multiple sets of input data, where each set of input data includes the camera's identification information and a time period, and different sets of input data include different time periods; input each set of input data into the prediction model to obtain the predicted number of target objects output by the prediction model for each set of input data; The time period included in the input data corresponding to the predicted number of target objects greater than the second threshold is used as the effective detection time period corresponding to the camera.

[0007] In combination with the first aspect, in one implementation, the method for screening images to be detected further includes: Obtain video data collected by some cameras during the verification period; The target detection model is used to perform target detection on the video data collected by each camera in the verification time period of some cameras, and the number of target objects detected by each camera in the verification time period is obtained; Calculate the absolute value of the difference between the predicted number of target objects and the detected number of target objects corresponding to each camera in the verification time period; If the number of absolute values ​​of the differences greater than the third threshold is greater than the fourth threshold, the effective detection time period corresponding to each camera is set as a default time period.

[0008] In combination with the first aspect, in one implementation, the method for screening images to be detected further includes: At a preset moment, a prediction model is obtained based on a training data set, where the training data set is constructed based on video data collected by each camera during a preset period of time before the preset moment.

[0009] In combination with the first aspect, in one implementation, the step of training a prediction model based on a training data set includes: Get N sets of training data from the training data set, where N is a positive integer. A set of training data includes a set of sample input data and a true value data. A set of sample input data includes camera C i The identity information and a time period T k , a true value data is obtained by the target detection model for camera C i In time period T k The number of target objects determined by target detection of the collected video data; Inputting N groups of sample input data into a preset neural network model to obtain output results of the preset neural network model for the N groups of sample input data; Calculate the loss based on the output result and N true value data; Check if the loss converges; If the loss has not converged, the parameters of the preset neural network model are updated, and the step of obtaining N sets of training data from the training data set is returned; If the loss converges, the current preset neural network model is used as the prediction model.

[0010] In combination with the first aspect, in one implementation, the identity information of the camera includes location information of the camera.

[0011] In combination with the first aspect, in one implementation, after detecting, for each channel of video data, whether a difference operation result of adjacent image frames in the video data is greater than a first threshold, the method further includes: If it is greater than the first threshold, detecting whether there is a target object in the image frame acquired later in the adjacent image frames by using a lightweight detection model; If the target object exists, the image frame acquired later in the adjacent image frames is used as the image to be detected.

[0012] In a second aspect, an embodiment of the present application provides a device for screening images to be detected, the device for screening images to be detected comprising: A determination module, used to determine the effective detection time period corresponding to each camera; An acquisition module is used to acquire the video data collected by each camera during its corresponding effective detection time period; A detection module, used for detecting, for each channel of video data, whether a difference operation result of adjacent image frames in the video data is greater than a first threshold; The screening module is used to use the image frame collected later in the adjacent image frames as the image to be detected if the value is greater than a first threshold.

[0013] In the third aspect, an embodiment of the present application provides an image screening device to be detected, wherein the image screening device to be detected comprises a processor, a memory, and an image screening program to be detected stored in the memory and executable by the processor, wherein when the image screening program to be detected is executed by the processor, the steps of the image screening method to be detected as described in the first aspect are implemented.

[0014] In a fourth aspect, an embodiment of the present application provides a computer-readable storage medium, on which a program for screening images to be detected is stored. When the program for screening images to be detected is executed by a processor, the steps of the method for screening images to be detected as described in the first aspect are implemented.

[0015] The beneficial effects brought by the technical solution provided by the embodiments of the present application include: In the embodiment of the present application, the effective detection time period corresponding to each camera is determined; the video data collected by each camera in its corresponding effective detection time period is obtained; for each channel of video data, whether the difference operation result of adjacent image frames in the video data is greater than a first threshold value is detected; if it is greater than the first threshold value, the image frame collected later in the adjacent image frames is used as the image to be detected. Through the embodiment of the present application, based on the video data collected by each camera in its corresponding effective detection time period, the frame difference algorithm is combined to screen the image to be detected for subsequent detection tasks, which reduces the amount of data faced by the detection task, thereby saving computing resources. BRIEF DESCRIPTION OF THE DRAWINGS

[0016] Figure 1 This is a flow chart of an embodiment of the method for screening images to be detected of the present application; Figure 2 for Figure 1 A detailed flow chart of step S10; Figure 3 A schematic diagram of a flow chart of another embodiment of the method for screening images to be detected of the present application; Figure 4 This is a schematic diagram of the functional modules of an embodiment of the device for screening images to be detected of the present application; Figure 5 Schematic diagram of the hardware structure of the image screening device to be detected involved in the embodiment of the present application. DETAILED DESCRIPTION

[0017] In order to enable those skilled in the art to better understand the solution of the present application, the technical solution in the embodiments of the present application will be clearly and completely described below in conjunction with the drawings in the embodiments of the present application. Obviously, the described embodiments are only part of the embodiments of the present application, not all of the embodiments. Based on the embodiments in the present application, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of this application.

[0018] In order to make the objectives, technical solutions and advantages of the present application clearer, the implementation methods of the present application will be further described in detail below with reference to the accompanying drawings.

[0019] In a first aspect, an embodiment of the present application provides a method for screening images to be detected.

[0020] In one embodiment, referring to Figure 1 , Figure 1 This is a flow chart of an embodiment of the method for screening images to be detected in this application. Figure 1 As shown, the method for screening images to be detected includes: Step S10, determining the effective detection time period corresponding to each camera; In this embodiment, it is assumed that the image to be detected is used for the task of detecting an electric vehicle entering an elevator. Considering that the event of an electric vehicle entering an elevator often occurs in a fixed time period, such as 6:00-9:00 and 18:00-21:00, the effective detection time period corresponding to each camera can be set to 6:00-9:00 and 18:00-21:00.

[0021] It should be noted that the effective detection time period is related to the detection task, and the effective detection time period corresponding to each camera can be flexibly set according to different detection tasks.

[0022] Further, in one embodiment, referring to Figure 2 , Figure 2 for Figure 1 Detailed flow chart of step S10 in FIG. Figure 2 As shown, step S10 includes: Step S101, constructing multiple groups of input data for each camera, wherein each group of input data includes identification information of the camera and a time period, and different groups of input data include different time periods; In this embodiment, taking camera 1 as an example, the constructed input data includes: Input data 1: Camera 1 identification information 1 + time period 1 (00:00~00:10); Input data 2: Camera 1’s identification information 1 + time period 2 (00:10~00:20); Input data 3: Camera 1’s identification information 1 + time period 3 (00:20~00:30); ...... Input data 144: identification information 1 of camera 1 + time period 144 (23:50~24:00).

[0023] As shown above, 24 hours are divided into 144 time periods at intervals of 10 minutes, and 144 sets of input data are constructed.

[0024] It is easy to understand that for each camera, the input data is constructed in the above manner, so that 144 sets of input data corresponding to each camera can be obtained.

[0025] It should be noted that the 10 minutes mentioned above is only an illustrative description, and the interval can also be 5 minutes, 20 minutes, 30 minutes, etc., which can be selected according to actual needs. Step S102, input each group of input data into the prediction model respectively, and obtain the predicted number of target objects output by the prediction model for each group of input data; Step S103: taking the time period included in the input data corresponding to the predicted number of target objects greater than the second threshold as the effective detection time period corresponding to the camera.

[0026] In this embodiment, taking the determination of the effective detection time period corresponding to camera 1 as an example, the 144 sets of input data corresponding to camera 1 are respectively input into the prediction model to obtain the predicted number of target objects output by the prediction model for each set of input data. For example, the predicted number of target objects corresponding to input data 1 is Y1, the predicted number of target objects corresponding to input data 2 is Y2, ..., and the predicted number of target objects corresponding to input data 144 is Y144.

[0027] Among them, if Y1, Y3, and Y7 are greater than the second threshold, the time period included in input data 1 (00:00~00:10), the time period included in input data 3 (00:20~00:30), and the time period included in input data 7 (01:00~01:10) are used as the effective detection time period corresponding to camera 1.

[0028] By analogy, the effective detection time period corresponding to each camera can be determined.

[0029] Among them, the prediction model is obtained through training.

[0030] Furthermore, in one embodiment, the method for screening images to be detected further includes: At a preset moment, a prediction model is obtained based on a training data set, where the training data set is constructed based on video data collected by each camera during a preset period of time before the preset moment.

[0031] In this embodiment, the preset time and the preset period are set according to actual needs. For example, the preset time is 00:00 every day, and the preset period is 10 days, then the training data set is constructed based on the video data collected by each camera 10 days before 00:00 every day.

[0032] Furthermore, in one embodiment, the prediction model is obtained based on the training data set, including: Step S00, obtaining N sets of training data from the training data set, wherein N is a positive integer, a set of training data includes a set of sample input data and a true value data, a set of sample input data includes camera C i The identity information and a time period T k , a true value data is obtained by the target detection model for camera C i In time period T k The number of target objects determined by target detection of the collected video data; In this embodiment, assuming that the image to be detected is used for the task of detecting an electric vehicle entering an elevator, the electric vehicle and a person can be used as target objects. Assuming that the cameras involved include camera 1 to camera 100, a training data set is constructed based on the video data collected by camera 1 to camera 100 in a preset period before a preset time.

[0033] For ease of description, a day is divided into 12 time periods at intervals of 1 hour, the preset time is 0:00 every day, and the preset time period is one day. If the preset time is 00:00 on September 10, the training data set includes: The following training data is constructed based on the video data captured by camera C1 from 00:00 to 24:00 on September 9: Training data C 11 :The identification information of camera C1 + time period T1 (00:00~01:00 on September 9) + the number of target objects H determined by target detection of the video data collected by camera C1 in time period T1 through the target detection model C1 .

[0034] Training data C12 :The identification information of camera C1 + time period T2 (01:00-02:00 on September 9) + the number of target objects H determined by target detection of the video data collected by camera C1 in time period T2 through the target detection model C2 . ...... Training data C 112 : Camera C1’s identity information + time period T 12 (23:00~24:00 on September 9) + Use the target detection model to detect camera C1 in time period T 12 The number of target objects H determined by target detection of the collected video data C12 .

[0036] By analogy, the training data constructed based on the video data captured by other cameras from 00:00 to 24:00 on September 9 will not be described in detail.

[0037] In summary, the training data set can be constructed.

[0038] Step S01, inputting N groups of sample input data into a preset neural network model to obtain output results of the preset neural network model for the N groups of sample input data; In this embodiment, the preset neural network model uses LSTM, but it can also be other types of network models, which are not limited here.

[0039] By inputting N groups of sample input data into the preset neural network model, the output results of the preset neural network model for each group of sample input data can be obtained, with a total of N output results.

[0040] Step S02, calculating the loss based on the output result and N true value data; In this embodiment, a suitable loss function is selected according to actual needs to calculate the loss based on N output results and N true value data.

[0041] Step S03, detecting whether the loss has converged; Step S04, if the loss has not converged, updating the parameters of the preset neural network model, and returning to the step of obtaining N groups of training data from the training data set; Step S05: if the loss converges, the current preset neural network model is used as the prediction model.

[0042] In this embodiment, if the loss has not converged, the parameters of the preset neural network model are updated, and then steps S00 to S03 are re-executed.

[0043] If the loss converges, the current preset neural network model is used as the prediction model.

[0044] Step S20, obtaining the video data collected by each camera during its corresponding effective detection time period; In this embodiment, the possibility of the target object appearing in the video data collected by each camera in time periods other than its corresponding effective detection time period is very low. Even if it is not used for the detection task, it has little impact on the final result of the detection task. Therefore, in order to save computing resources, only the video data collected by each camera in its corresponding effective detection time period is obtained.

[0045] Step S30, for each channel of video data, detecting whether a difference operation result of adjacent image frames in the video data is greater than a first threshold; Step S40: If the value is greater than the first threshold, the image frame collected later in the adjacent image frames is used as the image to be detected.

[0046] In this embodiment, it is easy to understand that one channel of video data is composed of continuous image frames. If each frame of image contained in each channel of video data is used for the detection task, it will lead to repeated detection of adjacent image frames whose monitoring screen has not changed, and will also cause a waste of computing resources. Therefore, for each channel of video data, a differential operation is performed on adjacent image frames in the video data, and the differential operation result is used to characterize the degree of difference between adjacent image frames. If the differential operation result is greater than the first threshold, it means that the monitoring screen has changed, and the image frame collected later in the adjacent image frames is used as the image to be detected for subsequent detection tasks.

[0047] In the embodiment of the present application, the effective detection time period corresponding to each camera is determined; the video data collected by each camera in its corresponding effective detection time period is obtained; for each channel of video data, whether the difference operation result of adjacent image frames in the video data is greater than a first threshold value is detected; if it is greater than the first threshold value, the image frame collected later in the adjacent image frames is used as the image to be detected. Through the embodiment of the present application, based on the video data collected by each camera in its corresponding effective detection time period, the frame difference algorithm is combined to screen the image to be detected for subsequent detection tasks, which reduces the amount of data faced by the detection task, thereby saving computing resources.

[0048] Furthermore, in one embodiment, the method for screening images to be detected further includes: Obtain video data collected by some cameras during the verification period; In this embodiment, the number of some cameras is set according to actual needs, for example, some cameras are one quarter of all cameras; the verification time is set according to actual needs.

[0049] Performing target detection on the video data collected by each camera in the verification time period of some cameras through the target detection model, obtaining the number of target objects detected by each camera in the verification time period; calculating the absolute value of the difference between the predicted number of target objects and the detected number of target objects corresponding to each camera in the verification time period; In this embodiment, taking one of the cameras as an example, the target detection model is used to perform target detection on the video data collected by the camera during the verification time period to obtain the number of target objects detected by the camera during the verification time period a. The identification information of the camera and the verification time period are input into the prediction model as input data to obtain the predicted number of target objects b corresponding to the camera during the verification time period.

[0050] Calculate the absolute value of the difference between a and b.

[0051] Assuming that there are 100 partial cameras, 100 absolute values ​​of differences will be obtained.

[0052] If the number of absolute values ​​of the differences greater than the third threshold is greater than the fourth threshold, the effective detection time period corresponding to each camera is set as a default time period.

[0053] In this embodiment, each difference absolute value is compared with the third threshold value to determine the number of difference absolute values ​​greater than the third threshold value. If the number of difference absolute values ​​greater than the third threshold value is greater than the fourth threshold value, it means that the prediction result of the prediction model is significantly different from the actual result, indicating that the prediction result of the prediction model is unreliable, and therefore the effective detection time period corresponding to the camera cannot be determined based on the prediction result of the prediction model, and the effective detection time period corresponding to each camera is set as the default time period. The default time period is set according to actual needs.

[0054] Furthermore, in one embodiment, the identity information of the camera includes location information of the camera.

[0055] In this embodiment, the position information of the camera may be the longitude and latitude coordinates of the camera. Experiments show that using the longitude and latitude coordinates of the camera as the identity information of the camera has a better training effect.

[0056] Further, in one embodiment, referring to Figure 3 , Figure 3 This is a flow chart of another embodiment of the method for screening images to be detected in the present application. Figure 3 As shown, after step S30, the following further includes: Step S50: if the value is greater than the first threshold, detecting whether the target object exists in the image frame acquired later in the adjacent image frames by using a lightweight detection model; Step S60: If the target object exists, the image frame collected later in the adjacent image frames is used as the image to be detected.

[0057] In this embodiment, it is easy to understand that, for example, if the final detection task is electric vehicle entering elevator detection, electric vehicles and people will be involved. Therefore, it is meaningful to use electric vehicles and people as images to be detected for subsequent detection tasks only when there are target objects such as electric vehicles in the image frame. Based on this, in order to further reduce the amount of tasks faced by the detection task, when the difference operation result of adjacent image frames is greater than the first threshold, the lightweight detection model is further used to detect whether there is a target object in the image frame collected later in the adjacent image frames. Only when there is a target object, the image frame collected later in the adjacent image frames is used as the image to be detected.

[0058] Among them, lightweight detection models can use NanoDet, YOLO, etc. This is only an example and does not constitute a limitation on lightweight detection models.

[0059] In a second aspect, an embodiment of the present application also provides a device for screening images to be detected.

[0060] In one embodiment, referring to Figure 4 , Figure 4 This is a functional module diagram of an embodiment of the image screening device to be detected in the present application. Figure 4 As shown, the image screening device to be detected includes: A determination module 10, used to determine an effective detection time period corresponding to each camera; An acquisition module 20 is used to acquire video data collected by each camera during its corresponding effective detection time period; The detection module 30 is used to detect, for each channel of video data, whether a difference operation result of adjacent image frames in the video data is greater than a first threshold; The screening module 40 is configured to use an image frame collected later in adjacent image frames as an image to be detected if the value is greater than a first threshold.

[0061] Furthermore, in one embodiment, the determination module 10 is used to: For each camera, construct multiple sets of input data, where each set of input data includes the camera's identification information and a time period, and different sets of input data include different time periods; input each set of input data into the prediction model to obtain the predicted number of target objects output by the prediction model for each set of input data; The time period included in the input data corresponding to the predicted number of target objects greater than the second threshold is used as the effective detection time period corresponding to the camera.

[0062] Furthermore, in one embodiment, the device for screening images to be detected further includes a reset module, which is used to: Obtain video data collected by some cameras during the verification period; The target detection model is used to perform target detection on the video data collected by each camera in the verification time period of some cameras, and the number of target objects detected by each camera in the verification time period is obtained; Calculate the absolute value of the difference between the predicted number of target objects and the detected number of target objects corresponding to each camera in the verification time period; If the number of absolute values ​​of the differences greater than the third threshold is greater than the fourth threshold, the effective detection time period corresponding to each camera is set as a default time period.

[0063] Furthermore, in one embodiment, the apparatus for screening images to be detected further includes a training module for: At a preset moment, a prediction model is obtained based on a training data set, where the training data set is constructed based on video data collected by each camera during a preset period of time before the preset moment.

[0064] Furthermore, in one embodiment, the training module is used to: Get N sets of training data from the training data set, where N is a positive integer. A set of training data includes a set of sample input data and a true value data. A set of sample input data includes camera C i The identity information and a time period T k , a true value data is obtained by the target detection model for camera C i In time period T k The number of target objects determined by target detection of the collected video data; Inputting N groups of sample input data into a preset neural network model to obtain output results of the preset neural network model for the N groups of sample input data; Calculate the loss based on the output result and N true value data; Check if the loss converges; If the loss has not converged, the parameters of the preset neural network model are updated, and the step of obtaining N sets of training data from the training data set is returned; If the loss converges, the current preset neural network model is used as the prediction model.

[0065] Furthermore, in one embodiment, the identity information of the camera includes location information of the camera.

[0066] Furthermore, in one embodiment, the screening module 40 is further configured to: If it is greater than the first threshold, detecting whether there is a target object in the image frame acquired later in the adjacent image frames by using a lightweight detection model; If the target object exists, the image frame acquired later in the adjacent image frames is used as the image to be detected.

[0067] Among them, the functional implementation of each module in the above-mentioned image screening device to be detected corresponds to the various steps in the above-mentioned image screening method embodiment to be detected, and its functions and implementation processes will not be repeated here one by one.

[0068] In a third aspect, an embodiment of the present application provides an image screening device to be detected, which may be a personal computer (PC), a laptop computer, a server, or other device with data processing capabilities.

[0069] Reference Figure 5 , Figure 5 Schematic diagram of the hardware structure of the image screening device to be detected involved in the embodiment of the present application. In the embodiment of the present application, the image screening device to be detected may include a processor, a memory, a communication interface and a communication bus.

[0070] The communication bus may be of any type and is used to interconnect the processor, the memory, and the communication interface.

[0071] The communication interface includes input / output (I / O) interface, physical interface and logical interface, etc., which are used to realize the interconnection of devices inside the image screening device to be detected, and the interface used to realize the interconnection of the image screening device to be detected with other devices (such as other computing devices or user devices). The physical interface can be an Ethernet interface, a fiber optic interface, an ATM interface, etc.; the user device can be a display, a keyboard, etc.

[0072] The memory can be various types of storage media, such as random access memory (RAM), read-only memory (ROM), non-volatile RAM (NVRAM), flash memory, optical storage, hard disk, programmable ROM (PROM), erasable PROM (EPROM), electrically erasable PROM (EEPROM), etc.

[0073] The processor may be a general-purpose processor, which may call the image screening program to be detected stored in the memory and execute the image screening method to be detected provided in the embodiment of the present application. For example, the general-purpose processor may be a central processing unit (CPU). The method executed when the image screening program to be detected is called may refer to the various embodiments of the image screening method to be detected of the present application, which will not be described in detail here.

[0074] Those skilled in the art will understand that Figure 5 The hardware structure shown in the figure does not constitute a limitation on the present application, and may include more or less components than shown in the figure, or combine certain components, or arrange the components differently.

[0075] In a fourth aspect, an embodiment of the present application also provides a computer-readable storage medium.

[0076] The computer-readable storage medium of the present application stores a program for screening images to be detected, wherein when the program for screening images to be detected is executed by a processor, the steps of the method for screening images to be detected as described above are implemented.

[0077] Among them, the method implemented when the image screening program to be detected is executed can refer to the various embodiments of the image screening method to be detected in this application, and will not be repeated here.

[0078] It should be noted that the serial numbers of the above-mentioned embodiments of the present application are only for description and do not represent the advantages or disadvantages of the embodiments.

[0079] The terms "including" and "having" and any variations thereof in the specification and claims of this application and the above-mentioned drawings are intended to cover non-exclusive inclusions. For example, a process, method, system, product or device that includes a series of steps or units is not limited to the listed steps or units, but optionally includes steps or units that are not listed, or optionally includes other steps or units inherent to these processes, methods, products or devices. The terms "first", "second" and "third" are used to distinguish different objects, etc., and do not represent a sequence, nor do they limit "first", "second" and "third" to different types.

[0080] In the description of the embodiments of the present application, "exemplary", "for example" or "for example" are used to indicate examples, illustrations or descriptions. Any embodiment or design described as "exemplary", "for example" or "for example" in the embodiments of the present application should not be interpreted as being more preferred or more advantageous than other embodiments or designs. Specifically, the use of words such as "exemplary", "for example" or "for example" is intended to present related concepts in a specific way.

[0081] In the description of the embodiments of the present application, unless otherwise specified, “ / ” means or, for example, A / B can mean A or B; the “and / or” in the text is merely a description of the association relationship of associated objects, indicating that three relationships may exist, for example, A and / or B can mean: A exists alone, A and B exist at the same time, and B exists alone. In addition, in the description of the embodiments of the present application, “multiple” refers to two or more than two.

[0082] In some processes described in the embodiments of the present application, multiple operations or steps that appear in a specific order are included, but it should be understood that these operations or steps may not be executed in the order in which they appear in the embodiments of the present application or in parallel, and the sequence number of the operation is only used to distinguish the different operations, and the sequence number itself does not represent any execution order. In addition, these processes may include more or fewer operations, and these operations or steps may be executed in sequence or in parallel, and these operations or steps may be combined.

[0083] Through the description of the above implementation methods, those skilled in the art can clearly understand that the above-mentioned embodiment methods can be implemented by means of software plus a necessary general hardware platform, and of course by hardware, but in many cases the former is a better implementation method. Based on such an understanding, the technical solution of the present application is essentially or the part that contributes to the prior art can be embodied in the form of a software product, which is stored in a storage medium (such as ROM / RAM, disk, CD) as described above, and includes a number of instructions for a terminal device to execute the methods described in each embodiment of the present application.

[0084] The above are only preferred embodiments of the present application, and are not intended to 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 applied in other related technical fields, are also included in the patent protection scope of the present application.

Claims

1. A method for screening images to be detected, characterized in that: The method for screening images to be detected comprises: Determine the effective detection time period corresponding to each camera; Obtain the video data collected by each camera during its corresponding effective detection time period; For each channel of video data, detecting whether a difference operation result of adjacent image frames in the video data is greater than a first threshold; If it is greater than the first threshold, the image frame acquired later in the adjacent image frames is used as the image to be detected.

2. The method for screening images to be detected as claimed in claim 1, characterized in that: Determining the effective detection time period corresponding to each camera includes: For each camera, construct multiple sets of input data, where each set of input data includes the camera's identification information and a time period, and different sets of input data include different time periods; input each set of input data into the prediction model to obtain the predicted number of target objects output by the prediction model for each set of input data; The time period included in the input data corresponding to the predicted number of target objects greater than the second threshold is used as the effective detection time period corresponding to the camera.

3. The method for screening images to be detected as claimed in claim 2, characterized in that: The method for screening images to be detected also includes: Obtain video data collected by some cameras during the verification period; The target detection model is used to perform target detection on the video data collected by each camera in the verification time period of some cameras, and the number of target objects detected by each camera in the verification time period is obtained; Calculate the absolute value of the difference between the predicted number of target objects and the detected number of target objects corresponding to each camera in the verification time period; If the number of absolute values ​​of the differences greater than the third threshold is greater than the fourth threshold, the effective detection time period corresponding to each camera is set as a default time period.

4. The method for screening images to be detected as claimed in claim 2, characterized in that: The method for screening images to be detected also includes: At a preset moment, a prediction model is obtained based on a training data set, where the training data set is constructed based on video data collected by each camera during a preset period of time before the preset moment.

5. The method for screening images to be detected as claimed in claim 4, characterized in that: The method of obtaining a prediction model based on the training data set includes: Get N sets of training data from the training data set, where N is a positive integer. A set of training data includes a set of sample input data and a true value data. A set of sample input data includes camera C i The identity information and a time period T k , a true value data is obtained by the target detection model for camera C i In time period T k The number of target objects determined by target detection of the collected video data; Inputting N groups of sample input data into a preset neural network model to obtain output results of the preset neural network model for the N groups of sample input data; Calculate the loss based on the output result and N true value data; Check if the loss converges; If the loss has not converged, the parameters of the preset neural network model are updated, and the step of obtaining N sets of training data from the training data set is returned; If the loss converges, the current preset neural network model is used as the prediction model.

6. The method for screening images to be detected as claimed in claim 5, characterized in that: The camera's identity information includes the camera's location information.

7. The method for screening images to be detected according to any one of claims 1 to 6, characterized in that: After detecting, for each channel of video data, whether a difference operation result of adjacent image frames in the video data is greater than a first threshold, the method further includes: If it is greater than the first threshold, detecting whether there is a target object in the image frame acquired later in the adjacent image frames by using a lightweight detection model; If the target object exists, the image frame acquired later in the adjacent image frames is used as the image to be detected.

8. A device for screening images to be detected, characterized in that: The image screening device to be detected comprises: A determination module, used to determine the effective detection time period corresponding to each camera; An acquisition module is used to acquire the video data collected by each camera during its corresponding effective detection time period; A detection module, used for detecting, for each channel of video data, whether a difference operation result of adjacent image frames in the video data is greater than a first threshold; The screening module is used to use the image frame collected later in the adjacent image frames as the image to be detected if the value is greater than the first threshold.

9. A device for screening images to be detected, characterized in that: The image screening device to be detected includes a processor, a memory, and an image screening program to be detected stored in the memory and executable by the processor, wherein when the image screening program to be detected is executed by the processor, the steps of the image screening method to be detected as described in any one of claims 1 to 7 are implemented.

10. A computer-readable storage medium, characterized in that: The computer-readable storage medium stores a program for screening images to be detected, wherein when the program for screening images to be detected is executed by a processor, the steps of the method for screening images to be detected as described in any one of claims 1 to 7 are implemented.

Images