Image content detection method, device, electronic device and storage medium

Abstract

The embodiment of the present invention provides an image content detection method, device, electronic device and storage medium, the method comprising: dividing the acquired image frame into regions to obtain S image sub-regions, where S is an integer greater than 1; performing suspected target detection on the S image sub-regions respectively to obtain suspected target detection results; when it is determined that there is an image position that meets a preset condition according to the suspected target detection results, intercepting an image partial region including the image position in the currently acquired image frame; and performing target detection on the image partial region. The embodiment of the present invention can reduce the power consumption of target detection.

Description

Technical Field

[0001] The present invention relates to the field of image processing technology, and in particular to an image content detection method, device, electronic device and storage medium. Background Art

[0002] With the development of image processing technology, image processing technology is becoming more and more widely used. Among them, object detection is currently a common image processing technology. At present, the object detection of an image is performed on each frame of the image and the entire area of ​​each frame of the image, which results in a relatively large power consumption of the object detection. Summary of the invention

[0003] Embodiments of the present invention provide an image content detection method, device, electronic device and storage medium to solve the problem of high power consumption in target detection.

[0004] In a first aspect, an embodiment of the present invention provides an image content detection method, comprising:

[0005] Dividing the acquired image frame into regions to obtain S image sub-regions, where S is an integer greater than 1;

[0006] Performing suspected target detection on the S image sub-regions respectively to obtain suspected target detection results;

[0007] When it is determined that there is an image position that meets a preset condition according to the suspected target detection result, intercepting a partial image area including the image position in the currently acquired image frame;

[0008] Performing object detection on the partial area of ​​the image.

[0009] In a second aspect, an embodiment of the present invention provides an image content detection device, comprising:

[0010] A division module, used for dividing the acquired image frame into regions to obtain S image sub-regions, where S is an integer greater than 1;

[0011] A first detection module, used to perform suspected target detection on the S image sub-regions respectively to obtain suspected target detection results;

[0012] A capture module, configured to capture a partial image region including the image position in the currently acquired image frame when it is determined that there is an image position that meets a preset condition according to the suspected target detection result;

[0013] The second detection module is used to perform target detection on the partial area of ​​the image.

[0014] In a third aspect, an embodiment of the present invention provides an electronic device, comprising: a memory, a processor, and a program or instruction stored in the memory and executable on the processor, wherein the program or instruction, when executed by the processor, implements the steps in the image content detection method provided in the embodiment of the present invention.

[0015] In a fourth aspect, an embodiment of the present invention provides a readable storage medium, characterized in that a program or instruction is stored on the readable storage medium, and when the program or instruction is executed by a processor, the steps in the image content detection method provided by the embodiment of the present invention are implemented.

[0016] In an embodiment of the present invention, the acquired image frame is divided into regions to obtain S image sub-regions, where S is an integer greater than 1; suspected target detection is performed on the S image sub-regions respectively to obtain suspected target detection results; when it is determined that there is an image position that meets a preset condition according to the suspected target detection result, a partial image region including the image position in the currently acquired image frame is intercepted; and target detection is performed on the partial image region. In this way, since suspected target detection is performed on each difference, target detection is then performed on the partial image region when the preset condition is met, the power consumption of target detection can be reduced. BRIEF DESCRIPTION OF THE DRAWINGS

[0017] Figure 1 is a flow chart of an image content detection method provided by an embodiment of the present invention;

[0018] Figure 2 is a schematic diagram of image division provided by an embodiment of the present invention;

[0019] Figure 3 is a structural diagram of an image content detection device provided by an embodiment of the present invention;

[0020] Figure 4 It is a structural diagram of an electronic device provided by an embodiment of the present invention. DETAILED DESCRIPTION

[0021] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are part of the embodiments of the present application, not all of the embodiments. Based on the embodiments in the present application, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of this application.

[0022] The terms "first", "second", etc. in the specification and claims of this application are used to distinguish similar objects, and are not used to describe a specific order or sequence. It should be understood that the terms used in this way can be interchangeable under appropriate circumstances, so that the embodiments of this application can be implemented in an order other than those illustrated or described here, and the objects distinguished by "first" and "second" are generally of the same type, and the number of objects is not limited. For example, the first object can be one or more.

[0023] See also Figure 1 , Figure 1 is a flow chart of an image content detection method provided by an embodiment of the present invention. Figure 1 As shown, the following steps are included:

[0024] Step 101: divide the acquired image frame into regions to obtain S image sub-regions, where S is an integer greater than 1.

[0025] The image frames acquired above may be image frames acquired by a camera, for example, images acquired by a built-in or external camera of an electronic device. Of course, this is not limited to this, for example, images may be received through a network, such as image frames of a received video. In addition, this acquisition may be receiving, or reading, image frames of a local video.

[0026] The above-mentioned area division may be performed according to pre-set division positions to obtain S areas, wherein S may be a pre-set integer greater than 1, for example, 2, 4 or 6, which may be specifically set according to the application scenario.

[0027] Step 102: Perform suspected target detection on the S image sub-regions respectively to obtain suspected target detection results.

[0028] The above-mentioned suspected target detection may be to detect whether there is a suspected target in the image sub-distinction, and does not perform specific detection on the target, for example: does not perform discrimination on the type, posture, behavior, action, attribute, etc. of the target.

[0029] Furthermore, the above-mentioned suspected target detection may also be defined as target detection with lower precision, accuracy and / or computational complexity than the target detection performed in step 104 .

[0030] The above-mentioned suspected target detection result may indicate that there are suspected targets in one or more image sub-regions, or may also indicate that there are no suspected targets in all image sub-regions for some image frames.

[0031] In addition, the above-mentioned image frames may be ultra-high resolution image frames or high resolution image frames. Of course, there is no limitation to this, and some low resolution image frames are also possible.

[0032] It should be noted that step 102 may be continuously executed, for example, step 102 is executed for each image frame captured by a camera or for consecutive image frames in a video.

[0033] Step 103: When it is determined that there is an image position that meets a preset condition according to the suspected target detection result, a partial image region including the image position in the currently acquired image frame is intercepted.

[0034] The preset condition may be that the score of an image pixel exceeds a preset threshold, so that the position of the pixel is used as the image position; or the preset condition may be that a suspected target is detected in the same image sub-region by multiple consecutive image frames, so that the image position corresponding to the suspected target is used as the image position. In the specific embodiment of the present invention, the preset condition is not limited, and may be pre-configured according to the application scenario or target detection requirements.

[0035] The above-mentioned intercepting of the image partial area including the image position in the currently acquired image frame may be intercepting the image frame with a preset width and height with the image position as the center.

[0036] The currently acquired image frame may be an image currently captured by a camera, or may be an image frame currently received or currently read.

[0037] Step 104: Perform object detection on the partial area of ​​the image.

[0038] The target detection may be to identify the type of the target, or may be to detect the relevant attributes of the target, or may be to detect the posture of the target, or may be to detect the behavior of the target, etc. In the embodiment of the present invention, target detection is not limited. Alternatively, the target detection may be to confirm the suspected target detected in step 102, that is, to confirm whether the suspected target is the target to be detected.

[0039] In the embodiment of the present invention, the above steps can be used to perform suspected target detection on each region, and then target detection is performed on a partial region of the image when preset conditions are met, thereby reducing power consumption of target detection.

[0040] It should be noted that the embodiments of the present invention can be applied to electronic devices, such as embedded devices, mobile phones, tablet computers, wearable devices, vehicles, etc., without limitation thereto.

[0041] As an optional implementation manner, there are overlapping areas between adjacent image sub-regions in the above S image sub-regions.

[0042] The overlapping area between adjacent image sub-regions may be that adjacent areas between adjacent image sub-regions are the overlapping areas.

[0043] For example, taking the above example of S being 4, assuming that the first image after video decoding is I, I is segmented into m rows and n columns of sub-regions. After segmentation, there are s = m × n image sub-regions, and there is an overlap of t pixels between each image sub-region. The parameters m, n, and t are all preset values ​​according to actual business needs. The overlap of the sub-regions after segmentation is as follows: Figure 2 shown.

[0044] In this implementation, since there are overlapping areas between adjacent image sub-regions, when each sub-region is detected, the continuity of the evaluation image content of each image sub-region can be guaranteed, thereby improving the accuracy of target detection.

[0045] As an optional implementation manner, when it is determined according to the suspected target detection result that there is an image position that meets a preset condition, intercepting an image partial area including the image position in the currently acquired image frame includes:

[0046] When the suspected target detection result indicates that there is a target sub-region of the suspected target, calculating scores of pixels in the target sub-region according to the location of the suspected target;

[0047] When the score of at least one pixel satisfies a preset condition, a partial image region including the at least one pixel in the currently acquired image frame is intercepted.

[0048] The suspected target may be an object with relevant attributes of the target, or an object with a high probability of being a target (such as higher than a preset threshold value), etc. The target may be pre-defined as an object to be detected, such as a specific person, a specific vehicle, a specific object, etc.

[0049] The score of the at least one pixel satisfies the preset condition may be that the score of the at least one pixel exceeds a preset threshold.

[0050] The above-mentioned currently acquired image frame may be an image frame acquired when an interception operation is performed when it is determined that the cumulative score of at least one element in the matrix exceeds a preset threshold, such as an image frame currently captured by the camera or an image frame currently read when the cumulative score of at least one element in the matrix exceeds the preset threshold, or may be an image frame captured or an image frame read by the camera after the cumulative score of at least one element in the matrix exceeds the preset threshold.

[0051] In this implementation, a partial area of ​​the image is intercepted based on the pixel scores to perform target detection, which can improve the accuracy of target detection.

[0052] Optionally, when the suspected target detection result indicates that there is a target sub-region of a suspected target, calculating the score of pixels in the target sub-region according to the location of the suspected target includes:

[0053] When the suspected target detection result indicates that there is a target sub-area of ​​the suspected target, the center position of the suspected target is determined, and the score of each pixel in the target sub-area is calculated based on the center position, and the pixel score is negatively correlated with the distance between the pixel and the center position.

[0054] In this implementation, a two-dimensional Gaussian distribution function may be used to determine the score of each pixel. For example, the area where the suspected target is located is the region of interest (ROI), and the center pixel position (x c ,y c ), which is the center position mentioned above, and other pixels are represented as pixels (i, j). Then the score of pixel (i, j) is obtained by the following formula:

[0055]

[0056] Among them, G(i,j) represents the score of pixel (i,j).

[0057] It should be noted that the above formula is only an example. For example, 0.5 in the above formula can be configured as other constants, or constants can be added to the above formula, etc., and there is no limitation on this.

[0058] Furthermore, the scores of the above center positions can be pre-configured or obtained by the above formula, such as pixel (i, j) is equal to pixel (x c ,y c ).

[0059] It represents a two-dimensional Gaussian distribution function that determines the decrease of pixel scores of an image sub-region from the center to the surrounding area.

[0060] It should be noted that the embodiments of the present invention are not limited to using a two-dimensional Gaussian distribution function to determine the score of each pixel. For example, the score can also be determined directly based on the distance between the pixel and the above-mentioned center position. For example, the distance between each pixel and the center position in the image sub-area is divided into multiple intervals, and each interval corresponds to a score.

[0061] In this implementation, the score of a pixel is negatively correlated with the distance between the pixel and the center position, which means that the farther the pixel is from the suspected target, the lower the score is, thereby reducing the triggering of the above preset conditions to further save power consumption.

[0062] It should also be noted that in the embodiment of the present invention, the score of each pixel in the target sub-area is not limited to being calculated based on the above-mentioned center position. For example, the score of each pixel in the target sub-area may also be directly calculated based on the area of ​​the suspected target.

[0063] In addition, the scores of pixels in the image sub-region where no suspected target is detected may be zero, or in the matrix implementation, the score accumulation operation is not performed for the image sub-region where no suspected target is detected.

[0064] Optionally, when the score of at least one pixel satisfies a preset condition, intercepting a partial image region including the at least one pixel in the currently acquired image frame comprises:

[0065] Accumulate the score of each pixel to the corresponding element of the matrix, wherein the number of elements included in the matrix is ​​the same as the number of pixels included in the acquired image frame, and the elements correspond to the pixels one by one;

[0066] When the cumulative score of at least one element in the matrix exceeds a preset threshold, a partial image region including at least one pixel in the currently acquired image frame is intercepted, wherein the at least one pixel corresponds one-to-one to the at least one element respectively.

[0067] For example, if the cumulative score of the reference element in the matrix exceeds a preset threshold, a partial image region including pixels corresponding one-to-one to the reference element in the currently acquired image frame may be intercepted.

[0068] The above matrix is ​​pre-defined, and the initial value of each pixel in the matrix may be zero.

[0069] It should be noted that, since the pixel frames or videos captured by the camera are continuous, the scores of the elements in the above matrix can be accumulated. Of course, in the embodiment of the present invention, the scores of the elements do not necessarily increase, and there may be a certain attenuation.

[0070] In this implementation, when the cumulative score of at least one element in the matrix exceeds a preset threshold, a partial image area including at least one pixel in the currently acquired image frame can be intercepted, thereby reducing the number of target detections and further saving power consumption.

[0071] Optionally, the cumulative score of the target element at the first moment is equal to the sum of the following two items:

[0072] The score of the pixel corresponding to the target element at the first moment;

[0073] a decay value of the cumulative score of the target element at a moment before the first moment;

[0074] The target element is any element in the matrix.

[0075] Among them, each moment corresponds to an image frame.

[0076] In this implementation, the current score of the pixel may be added to the attenuation value of the score at the previous moment to obtain the score at the current moment, thereby reducing the triggering of the preset conditions to further save power consumption.

[0077] Furthermore, due to the attenuation value of the cumulative score of the above-mentioned target element at the previous moment before the first moment, it can be achieved that some elements are attenuated to 0 without increasing the score. For example: a suspected target is not detected in a certain image sub-region for m consecutive frames, and thus these m frames have no scores, and the score of the corresponding element decays to 0.

[0078] In addition, the decay value of the above-mentioned cumulative score can use the exponential decay method to decay the score of the above-mentioned matrix (which can be expressed as M). Assume that if there is no new score increase, the score of each element in the matrix M decays to 0 after m frames, that is, each consecutive m frames are regarded as a time statistical period, and the matrix M in this time statistical period is decayed by the exponential decay method. The score of each element in the corresponding image frame (which can be expressed as the original image I) in the current cumulative matrix M is recorded as k x,y , then the element at the next moment (next frame) decays to multiplication by the preset coefficient α(t):

[0079] k x,y (t+1)=α(t)×k(t)

[0080] α(t)=exp[-θ×t]

[0081]

[0082] Among them, the above m and ε are constants, and the above θ formula can make the coefficient decay from 1 to close to 0.

[0083] Of course, in the embodiment of the present invention, the exponential decay method is not limited to decaying the score of the element. For example, the score of the element may be decayed by a certain percentage every other image frame, such as decaying by 20% every other image frame.

[0084] As an optional implementation manner, performing suspected target detection on the S image sub-regions respectively to obtain suspected target detection results includes:

[0085] Use the first network model to perform suspected target detection on the S image sub-regions respectively to obtain suspected target detection results;

[0086] The performing target detection on the partial area of ​​the image includes:

[0087] Performing target detection on the partial area of ​​the image using a second network model;

[0088] The computational complexity of the first network model is less than that of the second network model.

[0089] The above-mentioned first network model and second network model can be pre-trained. In addition, since the computational amount of the first network model is less than that of the second network model, the first network model can also be called a shallow target recognition network model, and the second network model can be called a deep target recognition network model.

[0090] In this implementation, the first network model is used to screen out suspected targets with less computational overhead, and the specific type of the target is not identified in detail, requiring only speed, not high recognition accuracy. The second network model is then used to perform target detection on a portion of the image.

[0091] In addition, the second network model can be a network that meets specific recognition accuracy according to actual business needs, such as a deep FasterRcnn network, and the first network model can be a shallow SSD network. Of course, there is no limitation to this, for example: it can also be other network models.

[0092] In this implementation, since only the second network model detects the captured image area after the above preset conditions are met, power consumption can be saved and computing efficiency can be improved.

[0093] In an embodiment of the present invention, the acquired image frame is divided into regions to obtain S image sub-regions, where S is an integer greater than 1; suspected target detection is performed on the S image sub-regions respectively to obtain suspected target detection results; when it is determined that there is an image position that meets a preset condition according to the suspected target detection result, a partial image region including the image position in the currently acquired image frame is intercepted; and target detection is performed on the partial image region. In this way, since suspected target detection is performed on each difference, target detection is then performed on the partial image region when the preset condition is met, the power consumption of target detection can be reduced.

[0094] See also Figure 3 , Figure 3 is a structural diagram of an image content detection device provided by an embodiment of the present invention, such as Figure 3 As shown, the image content detection device 300 includes:

[0095] A division module 301 is used to divide the acquired image frame into regions to obtain S image sub-regions, where S is an integer greater than 1;

[0096] A first detection module 302 is used to perform suspected target detection on the S image sub-regions respectively to obtain suspected target detection results;

[0097] The interception module 303 is used to intercept the image partial area including the image position in the currently acquired image frame when it is determined that there is an image position that meets the preset conditions according to the suspected target detection result;

[0098] The second detection module 304 is used to perform object detection on the partial area of ​​the image.

[0099] Optionally, the interception module includes:

[0100] a calculation unit, configured to calculate scores of pixels in the target subregion according to the location of the suspected target when the suspected target detection result indicates that there is a target subregion of the suspected target;

[0101] The interception unit is used to intercept the image partial area including the at least one pixel in the currently acquired image frame when the score of the at least one pixel meets the preset condition.

[0102] Optionally, the calculation unit is used to determine the center position of the suspected target when the suspected target detection result indicates that there is a target sub-area of ​​the suspected target, and calculate the score of each pixel in the target sub-area based on the center position, and the pixel score is negatively correlated with the distance between the pixel and the center position.

[0103] Optionally, the interception unit is used to accumulate the score of each pixel to the corresponding element of the matrix, wherein the number of elements included in the matrix is ​​the same as the number of pixels included in the acquired image frame, and the elements correspond to the pixels one-to-one; and when the cumulative score of at least one element in the matrix exceeds a preset threshold, intercepting a partial image area including at least one pixel in the currently acquired image frame, wherein the at least one pixel corresponds one-to-one to the at least one element, respectively.

[0104] Optionally, the cumulative score of the target element at the first moment is equal to the sum of the following two items:

[0105] The score of the pixel corresponding to the target element at the first moment;

[0106] a decay value of the cumulative score of the target element at a moment before the first moment;

[0107] The target element is any element in the matrix.

[0108] Optionally, the first detection module is used to use the first network model to perform suspected target detection on the S image sub-regions respectively to obtain suspected target detection results;

[0109] The second detection module is used to perform target detection on the partial area of ​​the image using a second network model;

[0110] The computational complexity of the first network model is less than that of the second network model.

[0111] The image content detection device provided by the embodiment of the present invention can achieve Figure 1 To avoid repetition, the various processes in the method embodiment are not described here.

[0112] It should be noted that the image content detection device in the embodiment of the present invention may be a device, or a component, an integrated circuit, or a chip in an electronic device.

[0113] See also Figure 4 , Figure 4 is a structural diagram of an electronic device provided by an embodiment of the present invention, such as Figure 4 As shown, the electronic device 400 includes: a memory 401, a processor 402, and a program or instruction stored in the memory 401 and executable on the processor 402, and the program or instruction, when executed by the processor 402, implements the steps in the above-mentioned image content detection method.

[0114] An embodiment of the present invention also provides a readable storage medium, on which a program or instruction is stored. When the program or instruction is executed by a processor, the various processes of the above-mentioned image content detection method embodiment are implemented, and the same technical effect can be achieved. To avoid repetition, it will not be repeated here.

[0115] It should be noted that, in this article, the terms "comprise", "include" or any other variant thereof are intended to cover non-exclusive inclusion, so that the process, method, article or device including a series of elements includes not only those elements, but also includes other elements not explicitly listed, or also includes elements inherent to such process, method, article or device. In the absence of further restrictions, the elements defined by the sentence "comprise one..." do not exclude the presence of other identical elements in the process, method, article or device including the element. In addition, it should be pointed out that the scope of the methods and devices in the embodiments of the present application is not limited to performing functions in the order shown or discussed, and may also include performing functions in a substantially simultaneous manner or in reverse order according to the functions involved, for example, the described method may be performed in an order different from that described, and various steps may also be added, omitted, or combined. In addition, the features described with reference to certain examples may be combined in other examples.

[0116] 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, 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, a magnetic disk, or an optical disk), and includes a number of instructions for a terminal (which can be a mobile phone, a computer, a server, an air conditioner, or a network device, etc.) to execute the methods described in each embodiment of the present application.

[0117] The embodiments of the present application are described above in conjunction with the accompanying drawings, but the present application is not limited to the above-mentioned specific implementation methods. The above-mentioned specific implementation methods are merely illustrative and not restrictive. Under the guidance of the present application, ordinary technicians in this field can also make many forms without departing from the purpose of the present application and the scope of protection of the claims, all of which are within the protection of the present application.

Claims

1. A method for detecting image content, characterized in that: include: Dividing the acquired image frame into regions to obtain S image sub-regions, where S is an integer greater than 1; Performing suspected target detection on the S image sub-regions respectively to obtain suspected target detection results; When it is determined that there is an image position that meets a preset condition according to the suspected target detection result, intercepting a partial image area including the image position in the currently acquired image frame, wherein the preset condition is that the score of the image pixel exceeds a preset threshold, or a plurality of consecutive image frames detect suspected targets in the same image sub-area; Performing target detection on the partial area of ​​the image; When it is determined according to the suspected target detection result that there is an image position that meets the preset condition, intercepting a partial image area including the image position in the currently acquired image frame, comprises: When the suspected target detection result indicates that there is a target sub-region of the suspected target, calculating scores of pixels in the target sub-region according to the location of the suspected target; When the score of at least one pixel satisfies a preset condition, intercepting a partial image region including the at least one pixel in the currently acquired image frame; When the score of at least one pixel satisfies a preset condition, intercepting a partial image region including the at least one pixel in the currently acquired image frame comprises: Accumulate the score of each pixel to the corresponding element of the matrix, wherein the number of elements included in the matrix is ​​the same as the number of pixels included in the acquired image frame, and the elements correspond to the pixels one by one; When the cumulative score of at least one element in the matrix exceeds a preset threshold, a partial image region including at least one pixel in the currently acquired image frame is intercepted, wherein the at least one pixel corresponds one-to-one to the at least one element respectively.

2. The method according to claim 1, characterized in that When the suspected target detection result indicates that there is a target sub-region of a suspected target, calculating the score of the pixels in the target sub-region according to the location of the suspected target, includes: When the suspected target detection result indicates that there is a target sub-area of ​​the suspected target, the center position of the suspected target is determined, and the score of each pixel in the target sub-area is calculated based on the center position, and the pixel score is negatively correlated with the distance between the pixel and the center position.

3. The method according to claim 1, characterized in that The cumulative score of the target element at the first moment is equal to the sum of the following two items: The score of the pixel corresponding to the target element at the first moment; a decay value of the cumulative score of the target element at a moment before the first moment; The target element is any element in the matrix.

4. The method according to any one of claims 1 to 3, characterized in that The performing suspected target detection on the S image sub-regions respectively to obtain suspected target detection results includes: Use the first network model to perform suspected target detection on the S image sub-regions respectively to obtain suspected target detection results; The performing target detection on the partial area of ​​the image includes: Performing target detection on the partial area of ​​the image using a second network model; The computational complexity of the first network model is less than that of the second network model.

5. An image content detection device, characterized in that: include: A division module, used for dividing the acquired image frame into regions to obtain S image sub-regions, where S is an integer greater than 1; A first detection module, used to perform suspected target detection on the S image sub-regions respectively to obtain suspected target detection results; A capture module, configured to capture a partial image region including an image position that satisfies a preset condition in a currently acquired image frame when it is determined that there is an image position that satisfies a preset condition according to the suspected target detection result, wherein the preset condition is that a score of an image pixel exceeds a preset threshold, or a suspected target is detected in the same image sub-region in multiple consecutive image frames; A second detection module, used to perform target detection on a partial area of ​​the image; The interception module comprises: a calculation unit, configured to calculate scores of pixels in the target subregion according to the location of the suspected target when the suspected target detection result indicates that there is a target subregion of the suspected target; A capture unit, configured to capture a partial image region including the at least one pixel in a currently acquired image frame when the score of the at least one pixel satisfies a preset condition; The interception unit is used to accumulate the score of each pixel to the corresponding element of the matrix, wherein the number of elements included in the matrix is ​​the same as the number of pixels included in the acquired image frame, and the elements correspond to the pixels one-to-one; and when the cumulative score of at least one element in the matrix exceeds a preset threshold, intercepting a partial image area including at least one pixel in the currently acquired image frame, wherein the at least one pixel corresponds to the at least one element respectively.

6. The device according to claim 5, characterized in that The calculation unit is used to determine the center position of the suspected target when the suspected target detection result indicates that there is a target sub-area of ​​the suspected target, and calculate the score of each pixel in the target sub-area based on the center position, and the pixel score is negatively correlated with the distance between the pixel and the center position.

7. The device according to claim 5, characterized in that The cumulative score of the target element at the first moment is equal to the sum of the following two items: The score of the pixel corresponding to the target element at the first moment; a decay value of the cumulative score of the target element at a moment before the first moment; The target element is any element in the matrix.

8. The device according to any one of claims 5 to 7, characterized in that The first detection module is used to use the first network model to perform suspected target detection on the S image sub-regions respectively to obtain suspected target detection results; The second detection module is used to perform target detection on the partial area of ​​the image using a second network model; The computational complexity of the first network model is less than that of the second network model.

9. An electronic device, characterized in that: include: A memory, a processor, and a program or instruction stored in the memory and executable on the processor, wherein the program or instruction, when executed by the processor, implements the steps in the image content detection method as claimed in any one of claims 1 to 4.

10. A readable storage medium, characterized in that: The readable storage medium stores a program or an instruction, and when the program or the instruction is executed by the processor, the steps in the image content detection method according to any one of claims 1 to 4 are implemented.

Images