Image detection method and device, nonvolatile storage medium and electronic equipment

Abstract

The application discloses an image detection method and device, a nonvolatile storage medium and an electronic device. The method comprises: obtaining a to-be-detected image of a to-be-detected space, wherein the to-be-detected image represents that no target object exists in the to-be-detected space; determining a time sequence image sequence based on the to-be-detected image, wherein the time sequence image sequence comprises: a plurality of preset detection images of the to-be-detected space, the plurality of preset detection images are time sequence adjacent, and the to-be-detected image is located at a middle position of the time sequence image sequence; in the case that the number of frames belonging to target detection images in the time sequence image sequence does not exceed a preset frame number threshold, determining that the to-be-detected image is an interest image, wherein the target detection image is a preset detection image in which the target object exists; and determining, as a residual object of the target object, a difference between the interest image and a calibration image. The application solves the technical problem of false positives in the existing residual object detection mode.

Description

Technical Field

[0001] This invention relates to the field of image recognition, and more specifically, to an image detection method, apparatus, non-volatile storage medium, and electronic device. Background Technology

[0002] Whether it's traditional banks, telecommunications operators, or emerging internet companies, all industries require large amounts of data centers to store servers and other hardware. Detecting and monitoring legacy items in data centers is crucial for ensuring unobstructed passageways and the stability and security of data center equipment.

[0003] Currently, the conventional method for detecting legacy items in data centers is through 24-hour dedicated personnel on duty, using monitoring equipment for manual review or regular on-site inspections. This method is costly in terms of manpower and prone to omissions due to staff fatigue. Another approach is to use deep learning models for legacy item detection, but this method is limited to detecting specific types of legacy items and is not suitable for large-scale deployment.

[0004] Furthermore, when monitoring equipment automatically detects abandoned items, false alarms may occur because the detected items may be being used by staff entering the computer room, leading to frequent false alarms.

[0005] There is currently no effective solution to the problem of false alarms in existing methods of detecting abandoned artifacts. Summary of the Invention

[0006] This invention provides an image detection method, apparatus, non-volatile storage medium, and electronic device to at least solve the technical problem of false alarms in existing methods for detecting unidentified objects.

[0007] According to one aspect of the present invention, an image detection method is provided, comprising: acquiring a target image of a target space, wherein the target image indicates that no target object exists in the target space; determining a temporal image sequence based on the target image, wherein the temporal image sequence includes: multiple frames of preset detection images of the target space, the multiple frames of the preset detection images being temporally adjacent, and the target image being located in the middle of the temporal image sequence; determining the target image as an interest image when the number of frames belonging to the target detection image in the temporal image sequence does not exceed a preset frame number threshold, wherein the target detection image is the preset detection image containing the target object; determining the difference between the interest image and a calibration image as a remnant of the target object, wherein the calibration image is pre-collected when neither the remnant nor the target object exists in the target space.

[0008] Optionally, obtaining a detection image in which no target object exists in the computer room includes: acquiring a preset detection image of the detection space at preset time intervals; detecting whether the target object exists in the preset detection image; and determining the preset detection image as the detection image if the target object does not exist in the preset detection image.

[0009] Optionally, determining the temporal image sequence based on the image to be detected includes: acquiring multiple frames of first detection images that are temporally adjacent to the image to be detected, wherein the multiple frames of first detection images are temporally adjacent, and there is a first detection image temporally adjacent to the image to be detected among the multiple frames of first detection images; acquiring multiple frames of second detection images that are temporally adjacent to the image to be detected, wherein the multiple frames of second detection images are temporally adjacent, and there is a second detection image temporally adjacent to the image to be detected among the multiple frames of second detection images; and determining multiple frames of the preset detection images in the temporal image sequence based on the multiple frames of first detection images, the image to be detected, and the multiple frames of second detection images.

[0010] Optionally, if the number of frames belonging to the target detection image in the time-series image sequence does not exceed a preset frame number threshold, determining the image to be detected as an image of interest includes: detecting whether the target object exists in each frame of the preset detection image in the time-series image sequence; determining that the preset detection image containing the target object is the target detection image; counting the number of frames of the target detection image in the time-series image sequence; and determining the image to be detected as the image of interest if the number of frames of the target detection image does not exceed the preset frame number threshold.

[0011] Optionally, determining the difference between the interest image and the calibration image as a remnant of the target object includes: dividing the interest image into multiple interest sub-images; determining a calibration sub-image corresponding to the interest sub-image in the calibration image, wherein the calibration image includes: multiple pre-divided calibration sub-images; determining the difference between the interest sub-image and the calibration sub-image; and stitching together the differences between the multiple interest sub-images and the calibration sub-images to determine the remnant.

[0012] Optionally, determining the difference between the interest sub-image and the calibration sub-image includes: determining a first grayscale mean of the calibration sub-image and a second grayscale mean of the interest sub-image; determining the grayscale difference between the first grayscale mean and the second grayscale mean; if the ratio of the grayscale difference to the first grayscale mean exceeds a preset grayscale threshold, determining the interest sub-image that differs from the calibration sub-image as a difference sub-image; and performing foreground extraction on the difference sub-image to determine the extracted foreground as the difference.

[0013] Optionally, foreground extraction is performed on the difference sub-image, and the extracted foreground is determined as the difference. This includes: based on a preset background model, setting the pixels of the same region between the difference sub-image and the calibration sub-image to a first preset value, and setting the pixels of the difference region between the difference sub-image and the calibration sub-image to a second preset value, wherein the preset background model is determined by initializing the calibration sub-image according to a running target detection algorithm, and the region in the difference sub-image where the pixels are set to the second preset value is the foreground; and extracting the region in the difference sub-image where the pixels are set to the second preset value as the difference.

[0014] According to another aspect of the present invention, an image detection apparatus is also provided, comprising: an acquisition module, configured to acquire a target image of a target space, wherein the target image indicates that no target object exists in the target space; a first determination module, configured to determine a temporal image sequence based on the target image, wherein the temporal image sequence includes: multiple frames of preset detection images of the target space, the multiple frames of the preset detection images being temporally adjacent, and the target image being located in the middle position of the temporal image sequence; a second determination module, configured to determine the target image as an interest image when the number of frames belonging to the target detection image in the temporal image sequence does not exceed a preset frame number threshold, wherein the target detection image is the preset detection image containing the target object; and a third determination module, configured to determine the difference between the interest image and the calibration image as a remnant of the target object, wherein the calibration image is pre-collected when neither the remnant nor the target object exists in the target space.

[0015] According to another aspect of the present invention, a non-volatile storage medium is also provided, the non-volatile storage medium being used to store a program, wherein the program controls the device where the non-volatile storage medium is located to execute the above-described image detection method during runtime.

[0016] According to another aspect of the present invention, an electronic device is also provided, comprising: a memory and a processor, the processor being configured to run a program stored in the processor, wherein the program executes the image detection method described above when it runs.

[0017] In this embodiment of the invention, a detection image of the detection space is acquired, wherein the detection image indicates that no target object exists in the detection space; a temporal image sequence is determined based on the detection image, wherein the temporal image sequence includes: multiple frames of preset detection images of the detection space, the multiple frames of preset detection images are temporally adjacent, and the detection image is located in the middle position of the temporal image sequence; if the number of frames belonging to the target detection image in the temporal image sequence does not exceed a preset frame number threshold, the detection image is determined to be an interest image, wherein the target detection image is a preset detection image containing a target object; the difference between the interest image and the calibration image is determined to be a remnant of the target object, wherein the calibration image is pre-collected under the condition that no remnant or target object exists in the detection space; thereby, the remnant of the target object can be detected even when the target object does not frequently appear in the detection space, ensuring that the detected remnant is an item left behind by the target object, and not an item used by the target object in the detection space, thus achieving the technical effect of false alarm detection results for remnant, and thus solving the technical problem of false alarm in existing remnant detection methods. Attached Figure Description

[0018] The accompanying drawings, which are included to provide a further understanding of the invention and form part of this application, illustrate exemplary embodiments of the invention and, together with their description, serve to explain the invention and do not constitute an undue limitation thereof. In the drawings:

[0019] Figure 1 This is a flowchart of an image detection method according to an embodiment of the present invention;

[0020] Figure 2 This is a schematic diagram of a low false alarm data center object detection method based on time-series personnel detection according to an embodiment of the present invention;

[0021] Figure 3 This is a schematic diagram of an image detection device according to an embodiment of the present invention;

[0022] Figure 4 This is a structural block diagram of a computer terminal according to an embodiment of the present invention. Detailed Implementation

[0023] To enable those skilled in the art to better understand the present invention, the technical solutions of the present invention will be clearly and completely described below with reference to the accompanying drawings of the embodiments of the present invention. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort should fall within the scope of protection of the present invention.

[0024] It should be noted that the terms "first," "second," etc., in the specification, claims, and accompanying drawings of this invention are used to distinguish similar objects and are not necessarily used to describe a specific order or sequence. It should be understood that such data can be interchanged where appropriate so that the embodiments of the invention described herein can be implemented in orders other than those illustrated or described herein. Furthermore, the terms "comprising" and "having," and any variations thereof, are intended to cover a non-exclusive inclusion; for example, a process, method, system, product, or apparatus that comprises a series of steps or units is not necessarily limited to those steps or units explicitly listed, but may include other steps or units not explicitly listed or inherent to such processes, methods, products, or apparatus.

[0025] According to an embodiment of the present invention, an image detection method embodiment is provided. It should be noted that the steps shown in the flowchart in the accompanying drawings can be executed in a computer system such as a set of computer-executable instructions. Furthermore, although a logical order is shown in the flowchart, in some cases, the steps shown or described may be executed in a different order than that shown here.

[0026] Figure 1 This is a flowchart of an image detection method according to an embodiment of the present invention, such as... Figure 1 As shown, the method includes the following steps:

[0027] Step S102: Obtain the image to be detected in the detection space, wherein the image to be detected indicates that there is no target object in the detection space;

[0028] Step S104: Determine a temporal image sequence based on the image to be detected, wherein the temporal image sequence includes: multiple frames of preset detection images in the space to be detected, the multiple frames of preset detection images are temporally adjacent, and the image to be detected is located in the middle position of the temporal image sequence.

[0029] Step S106: If the number of frames belonging to the target detection image in the time-series image sequence does not exceed a preset frame number threshold, the image to be detected is determined to be an interest image, wherein the target detection image is a preset detection image containing a target object;

[0030] Step S108: The difference between the interest image and the calibration image is determined as the residue of the target object, wherein the calibration image is pre-collected under the condition that there is no residue or target object in the space to be detected.

[0031] In this embodiment of the invention, a detection image of the detection space is acquired, wherein the detection image indicates that no target object exists in the detection space; a temporal image sequence is determined based on the detection image, wherein the temporal image sequence includes: multiple frames of preset detection images of the detection space, the multiple frames of preset detection images are temporally adjacent, and the detection image is located in the middle position of the temporal image sequence; if the number of frames belonging to the target detection image in the temporal image sequence does not exceed a preset frame number threshold, the detection image is determined to be an interest image, wherein the target detection image is a preset detection image containing a target object; the difference between the interest image and the calibration image is determined to be a remnant of the target object, wherein the calibration image is pre-collected under the condition that no remnant or target object exists in the detection space; thereby, the remnant of the target object can be detected even when the target object does not frequently appear in the detection space, ensuring that the detected remnant is an item left behind by the target object, and not an item used by the target object in the detection space, thus achieving the technical effect of false alarm detection results for remnant, and thus solving the technical problem of false alarm in existing remnant detection methods.

[0032] In step S102 above, the space to be tested can be a computer room, and the target object can be the staff of the computer room.

[0033] Optionally, if the target object is present in the space to be detected, the detected object is likely to be an item brought into the computer room by staff, such as a tool used by staff. Therefore, the detected object is a false alarm.

[0034] Optionally, if the target object is not found in the space to be detected, the detected object is likely to be an item left behind by staff in the computer room, and therefore the detected object will not be a false alarm.

[0035] In step S104 above, when detecting remnants based on the image to be detected, multiple frames of predicted detection images including the image to be detected can be obtained as a time-series image sequence, and the target object can be detected frequently in the space to be detected based on the time-series image sequence.

[0036] In step S106 above, by detecting the number of frames of target detection images containing target objects in the time-series image sequence, it is determined whether the target object is frequently active in the detection space. Then, if the target object is not frequently active in the detection space, the image to be detected is used as an image of interest for object detection. This avoids using non-images of interest for object detection, which would consume computing power for object detection and thus improves detection efficiency.

[0037] In step S108 above, the calibration image can be collected in the space to be detected under the condition that there are no leftover objects or target objects in advance. The calibration image can be used as a benchmark to determine whether there are leftover objects in the space to be detected. If there is a difference between the image of interest and the calibration image, the difference may be leftover objects.

[0038] As an optional embodiment, obtaining a detection image in which no target object exists in the computer room includes: acquiring a preset detection image of the space to be detected at preset time intervals; detecting whether a target object exists in the preset detection image; and determining the preset detection image as the detection image if no target object exists in the preset detection image.

[0039] In the above embodiments of the present invention, preset detection images of the space to be detected are collected at preset time intervals, which can continuously detect the space to be detected. Only when the preset detection image does not contain a target object will the preset detection image be used as the image to be detected and further object detection will be performed; otherwise, if the preset detection image contains a target object, further object detection will not be performed. This avoids using a preset detection image with a target object for object detection, which would result in the invalid use of object detection computing power and thus improve the efficiency of object detection.

[0040] Optionally, a preset detection image of the space to be detected can be acquired by a monitoring camera.

[0041] Optionally, detecting whether a target object exists in a preset detection image includes: performing target object detection on the preset detection image based on a deep learning model, wherein the deep learning model is obtained by pre-training using features containing the target object.

[0042] Optionally, the preset detection images are collected at preset time intervals, so multiple temporally adjacent preset detection images can be obtained.

[0043] As an optional embodiment, determining a temporal image sequence based on an image to be detected includes: acquiring multiple first detection images that are temporally preceding the image to be detected, wherein the multiple first detection images are temporally adjacent, and one of the multiple first detection images is a first detection image temporally adjacent to the image to be detected; acquiring multiple second detection images that are temporally following the image to be detected, wherein the multiple second detection images are temporally adjacent, and one of the multiple second detection images is a second detection image temporally adjacent to the image to be detected; and determining multiple preset detection images in the temporal image sequence based on the multiple first detection images, the image to be detected, and the multiple second detection images.

[0044] In the above embodiments of the present invention, during the process of determining a time-series image sequence based on an image to be detected, multiple frames of first detection images preceding the image to be detected and multiple frames of second detection images preceding the image to be detected can be obtained based on the image to be detected. Based on the multiple frames of first detection images, the image to be detected, and the multiple frames of second detection images, multiple frames of preset detection images in the time-series image sequence can be determined. Thus, based on the time-series image sequence, it is possible to detect whether the target object is frequently active within the time period indicated by the time-series image sequence.

[0045] Optionally, the temporal image sequence includes multiple frames of images to be detected collected at preset time intervals within the corresponding time period.

[0046] Optionally, the preset detection image can be a sequence of multiple frames of images collected at preset time intervals. Therefore, in the process of determining the time sequence of images based on the image to be detected, the image to be detected can be used as a reference to extract preset detection images of a preset number of frames forward and backward from the frame image sequence as the time sequence of images.

[0047] Optionally, the multiple first detection images are temporally adjacent first detection sequences, and the last first detection image in the first detection sequence is temporally adjacent to the image to be detected.

[0048] Optionally, the multiple frames of second detection images are temporally adjacent second detection sequences, and the first second detection image of the second detection sequence is temporally adjacent to the image to be detected.

[0049] Optionally, if the image to be detected is the last frame of the preset detection image collected at a preset time interval, since the second detection image has not been collected, the second detection image cannot be obtained. Therefore, the collection of preset detection images can continue at the preset time interval until the number of frames of the collected preset detection images that can be used as the second detection image meets the requirements of the time sequence image sequence. Then, the image of interest is determined based on the time sequence image sequence.

[0050] Optionally, the first and second detection images have the same number of frames in the temporal image sequence.

[0051] As an optional embodiment, determining the image to be detected as an image of interest when the number of frames belonging to the target detection image in the time-series image sequence does not exceed a preset frame number threshold includes: detecting whether a target object exists in each frame of the preset detection image in the time-series image sequence; determining the preset detection image with the target object as the target detection image; counting the number of frames of the target detection image in the time-series image sequence; and determining the image to be detected as an image of interest when the number of frames of the target detection image does not exceed the preset frame number threshold.

[0052] In the above embodiments of the present invention, the target detection image is a preset detection image containing a target object in a time-series image sequence. Therefore, by detecting the frame number of the target detection image in the time-series image sequence, it can be determined whether the target object is frequently active within the time period indicated by the time-series image sequence. If the target object is frequently active within the time period, the detection result based on the image to be detected may be an item that the target object still needs to use, rather than a relic. If the target object is not frequently active within the time period, it indicates that the target object has left the detection space. Therefore, the image to be detected is used as an image of interest for relic detection, which can ensure that the image of interest meets the requirements for relic detection, avoids the invalid occupation of computing power for relic detection, and thus improves the detection efficiency of relic.

[0053] As an optional embodiment, determining the difference between the interest image and the calibration image as the residue of the target object includes: dividing the interest image into multiple interest sub-images; determining the calibration sub-image corresponding to the interest sub-image in the calibration image, wherein the calibration image includes: multiple pre-divided calibration sub-images; determining the difference between the interest sub-image and the calibration sub-image; and stitching together the differences between the multiple interest sub-images and the calibration sub-image to determine the residue.

[0054] In the above embodiments of the present invention, when the differences between the interest image and the calibration image are determined, the interest image and the calibration image can be divided into multiple small sub-images, and then the differences are compared based on the divided sub-images, and then the images are stitched together. This can reduce the computational power required and improve the detection efficiency during the comparison of the differences between the interest image and the calibration image.

[0055] As an optional embodiment, determining the difference between the interest sub-image and the calibration sub-image includes: determining a first gray-scale mean of the calibration sub-image and a second gray-scale mean of the interest sub-image; determining the gray-scale difference between the first gray-scale mean and the second gray-scale mean; if the ratio of the gray-scale difference to the first gray-scale mean exceeds a preset gray-scale threshold, determining the interest sub-image that differs from the calibration sub-image as a difference sub-image; and performing foreground extraction on the difference sub-image to determine the extracted foreground as the difference.

[0056] In the above embodiments of the present invention, when the difference between the interest sub-image and the calibration sub-image is determined, the gray-scale mean values ​​of the interest sub-image and the calibration sub-image can be compared. If the difference between the gray-scale mean values ​​of the interest sub-image and the calibration sub-image is too large, it indicates that there is a significant difference between the interest sub-image and the calibration sub-image. This difference may be caused by residual objects. Therefore, the interest sub-image can be determined as a difference sub-image, and the residual objects in the difference sub-image can be obtained by performing foreground extraction on the difference sub-image.

[0057] It should be noted that if the mean grayscale values ​​of the interest submap and the calibration submap are different, but the difference between the mean grayscale values ​​of the interest submap and the calibration submap is not significant, it means that there is no obvious difference between the interest submap and the calibration submap. In this case, the difference may be caused by noise. Therefore, the residue of the target object cannot be obtained based on the interest submap, so the interest submap will not be used as a difference submap for residue extraction.

[0058] As an optional embodiment, foreground extraction from the difference sub-image and determining the extracted foreground as the difference includes: based on a preset background model, setting the pixels of the same region between the difference sub-image and the calibration sub-image to a first preset value, and setting the pixels of the difference region between the difference sub-image and the calibration sub-image to a second preset value, wherein the preset background model is determined by initializing the calibration sub-image according to the running target detection algorithm, and the region in the difference sub-image where the pixels are set to the second preset value is the foreground; extracting the region in the difference sub-image where the pixels are set to the second preset value as the difference.

[0059] In the above embodiments of the present invention, during the foreground extraction process of the difference sub-image, a preset background model can be pre-trained based on each calibration sub-image of the calibration image. The preset background model uses the content of the corresponding difference sub-image as the background and then compares the differences with the difference sub-image. Based on the preset background model, the pixels of the same area in the difference sub-image and the calibration sub-image can be set to a first preset value, and the pixels of the difference area between the difference sub-image and the calibration sub-image can be set to a second preset value. This achieves binarization of the difference sub-image, which can highlight the difference area based on the binarized difference sub-image, making the difference area the foreground, and facilitating the extraction of residual objects.

[0060] Optionally, the first preset value can be 0, and the second preset value can be 255.

[0061] The present invention also provides a preferred embodiment, which provides a low false alarm method for detecting abandoned items in a computer room based on time-series personnel detection.

[0062] Figure 2 This is a schematic diagram of a low-false-alarm data center debris detection method based on time-series personnel detection according to an embodiment of the present invention, as shown below. Figure 2 As shown, it includes: a segmented background modeling module, a personnel detection module, a time series analysis module, a segmented anomaly detection module, and a debris location module.

[0063] Optionally, a block background modeling module is used for block background modeling based on the model-free update Vibe algorithm. To improve the subsequent detection speed of remnants, the background image (i.e., the calibration image) is divided into multiple sub-regions (i.e., calibration sub-images). Considering the stable background characteristics of the computer room scene, the model-free update Vibe algorithm is used to independently model the background of each sub-region (i.e., the calibration sub-image). The specific modeling steps are as follows:

[0064] 1. Based on the data center monitoring, collect an RGB background image (such as a calibration image), and divide the RGB background image into N×N (such as 4×4) local sub-images (such as calibration sub-images).

[0065] 2. Each local subgraph (such as the calibration subgraph) is initialized using the Vibe algorithm to obtain the local background model B. i (e.g., a preset background model), where i∈[0,N 2 -1], and the background model will not be updated during the subsequent foreground detection based on the local background model.

[0066] 3. Extract the mean grayscale value m for each local background sub-image. i where i∈[0,N 2 -1].

[0067] Optionally, a personnel detection module is used for personnel detection based on a deep learning model. A personnel detection model is trained using a YOLOv5s network. This model is then used to detect the presence of personnel (such as target objects) in each image captured by the computer room monitoring system (e.g., a preset detection image). If a human body is detected, subsequent object detection is not performed; otherwise, subsequent steps for object detection continue.

[0068] Optionally, a time-series analysis module is used for filtering images of interest based on time-series analysis. If no people are present in the current image (e.g., the image to be detected), an image sequence (e.g., a time-series image sequence) is formed based on the time-series neighboring images of the current image (e.g., the image to be detected). Images of interest are then filtered based on the detection of people in the image sequence. The specific steps are as follows:

[0069] 1. If the graph I at time t t If no people are detected in the image to be detected, then create image I. t The corresponding time-series image sequence S t ={I t-n ,…,I t-1 ,I t ,I t+1 ,…,I t+n}, time-series image sequence S t It consists of the current image (such as the image to be detected), the previous n frames of the current image (such as multiple frames of the first detection image), and the next n frames of the current image (multiple frames of the second detection image).

[0070] It should be noted that in practical applications, one frame can be extracted per second to expand the temporal domain perception range of people nearby at the current moment.

[0071] 2. Set a preset frame rate threshold k to determine whether there is human activity near time t. If the preset image sequence S t If people are detected in k or more target detection images, it proves that there is still human activity near time t, and no further object detection is performed; otherwise, image I... t It can be used as an image of interest for subsequent object detection.

[0072] Optionally, the preset frame rate threshold k can be set to 3.

[0073] Optionally, a block-based anomaly detection module is used for anomaly foreground detection based on a local background model (such as a preset background model). The image of interest is divided into multiple interest sub-images. The mean grayscale value of each interest sub-image is compared with the mean grayscale value of the corresponding calibration sub-image to determine whether the current sub-image (e.g., the interest sub-image) needs foreground detection based on the corresponding local background model (such as a preset background model). The specific steps are as follows:

[0074] 1. Divide the image of interest into N×N interest sub-images, and extract the grayscale mean value for each interest sub-image. where i∈[0,N] 2 -1]; The following formula can be used to determine whether there is an abnormal foreground in the current subgraph, thus eliminating noise interference:

[0075]

[0076] Optionally, if the current interest subgraph satisfies the above formula, then based on the corresponding local background model B... i Detecting abnormal prospects.

[0077] 2. After dividing the interest image into N×N interest sub-images, corresponding to N×N calibration sub-images, the grayscale values ​​of the pixels corresponding to the foreground and background segmentation results of interest sub-images that do not satisfy the above formula (i.e., difference sub-images) are all assigned to 0. Interest sub-images that satisfy the above formula (i.e., difference sub-images) will be based on the corresponding local background model B. i The foreground and background segmentation results are obtained through the Vibe algorithm. The pixel with a pixel value of 255 is the suspected area of ​​the object, thus realizing the binarization of the interest sub-image.

[0078] Optionally, the remnant localization module is used to extract connected components with a pixel grayscale value of 255 from the binarized image obtained by the block anomaly detection module, fill the holes in the connected components to ensure the integrity of the remnant localization, and remove connected components whose circumscribed rectangles have a height and width both less than p pixels to further eliminate the influence of noise. The remaining connected components are the pixel coordinates of the remnant.

[0079] Alternatively, the object location module can be implemented using an OpenCV function, which is as follows: cv::fillHole(Mat foregroundMask, Mat foregroundMask).

[0080] Optionally, at an input image size of 1920×1080, connected components with a bounding rectangle height and width both less than 50 pixels are removed to further eliminate the influence of noise.

[0081] Alternatively, the OpenCV function and code flow can be used as follows:

[0082]

[0083]

[0084] Optionally, the remaining connected components are the pixel coordinates of the remaining objects.

[0085] The technical solution provided in this application allows for rapid algorithm deployment based on existing surveillance cameras in the computer room, resulting in low hardware costs. It also replaces 24 / 7 manual monitoring or inspection, saving labor costs. It eliminates the need for prior data collection on abandoned objects and has no restrictions on the type of abandoned objects, facilitating widespread application across different types of computer rooms. Addressing the characteristic that the primary moving targets in computer rooms are personnel, it locates abandoned objects based on time-series personnel detection results, significantly reducing false detections and improving the method's robustness. The proposed block-based background modeling and block-based anomaly detection scheme only detects abandoned objects in moving areas, greatly improving detection speed and achieving pixel-level localization of abandoned objects in the computer room. Therefore, it can utilize time-series personnel detection conditions for abandoned object detection while reducing labor costs, significantly decreasing false detections and reducing frequent false alarms caused by personnel entering the computer room.

[0086] According to an embodiment of the present invention, an image detection device embodiment is also provided. It should be noted that the image detection device can be used to execute the image detection method in the embodiment of the present invention, and the image detection method in the embodiment of the present invention can be executed in the image detection device.

[0087] Figure 3 This is a schematic diagram of an image detection device according to an embodiment of the present invention, such as... Figure 3As shown, the device may include: an acquisition module 32, used to acquire a detection image of a detection space, wherein the detection image indicates that no target object exists in the detection space; a first determination module 34, used to determine a temporal image sequence based on the detection image, wherein the temporal image sequence includes: multiple frames of preset detection images of the detection space, the multiple frames of preset detection images are temporally adjacent, and the detection image is located in the middle of the temporal image sequence; a second determination module 36, used to determine the detection image as an interest image when the number of frames belonging to the target detection image in the temporal image sequence does not exceed a preset frame number threshold, wherein the target detection image is a preset detection image containing a target object; and a third determination module 38, used to determine the difference between the interest image and the calibration image as a remnant of the target object, wherein the calibration image is pre-collected under the condition that no remnant or target object exists in the detection space.

[0088] It should be noted that the acquisition module 32 in this embodiment can be used to execute step S102 in this application embodiment, the first determination module 34 in this embodiment can be used to execute step S104 in this application embodiment, the second determination module 36 in this embodiment can be used to execute step S106 in this application embodiment, and the third determination module 38 in this embodiment can be used to execute step S108 in this application embodiment. The examples and application scenarios implemented by the above modules and corresponding steps are the same, but are not limited to the content disclosed in the above embodiments.

[0089] In this embodiment of the invention, a detection image of the detection space is acquired, wherein the detection image indicates that no target object exists in the detection space; a temporal image sequence is determined based on the detection image, wherein the temporal image sequence includes: multiple frames of preset detection images of the detection space, the multiple frames of preset detection images are temporally adjacent, and the detection image is located in the middle position of the temporal image sequence; if the number of frames belonging to the target detection image in the temporal image sequence does not exceed a preset frame number threshold, the detection image is determined to be an interest image, wherein the target detection image is a preset detection image containing a target object; the difference between the interest image and the calibration image is determined to be a remnant of the target object, wherein the calibration image is pre-collected under the condition that no remnant or target object exists in the detection space; thereby, the remnant of the target object can be detected even when the target object does not frequently appear in the detection space, ensuring that the detected remnant is an item left behind by the target object, and not an item used by the target object in the detection space, thus achieving the technical effect of false alarm detection results for remnant, and thus solving the technical problem of false alarm in existing remnant detection methods.

[0090] As an optional embodiment, the acquisition module includes: a acquisition unit, used to acquire a preset detection image of the space to be detected at preset time intervals; a first detection unit, used to detect whether a target object exists in the preset detection image; and a first determination unit, used to determine the preset detection image as the image to be detected when no target object exists in the preset detection image.

[0091] As an optional embodiment, the first determining module includes: a first acquiring unit, configured to acquire multiple frames of first detection images that are time-series preceding the image to be detected, wherein the multiple frames of first detection images are time-adjacent, and there is a first detection image that is time-adjacent to the image to be detected among the multiple frames of first detection images; a second acquiring unit, configured to acquire multiple frames of second detection images that are time-series following the image to be detected, wherein the multiple frames of second detection images are time-adjacent, and there is a second detection image that is time-adjacent to the image to be detected among the multiple frames of second detection images; and a second determining unit, configured to determine multiple preset detection images in the time-series image sequence based on the multiple frames of first detection images, the image to be detected, and the multiple frames of second detection images.

[0092] As an optional embodiment, the second determining module includes: a second detection unit, used to detect whether a target object exists in each frame of a preset detection image in a time-series image sequence; a third determining unit, used to determine that the preset detection image containing the target object is a target detection image; a statistics unit, used to count the number of frames of the target detection images in the time-series image sequence; and a fourth determining unit, used to determine that the image to be detected is an image of interest when the number of frames of the target detection images does not exceed a preset frame number threshold.

[0093] As an optional embodiment, the third determining module includes: a segmentation unit for dividing the interest image into multiple interest sub-images; a fifth determining unit for determining a calibration sub-image corresponding to the interest sub-image in the calibration image, wherein the calibration image includes: multiple pre-divided calibration sub-images; a sixth determining unit for determining the differences between the interest sub-images and the calibration sub-images; and a stitching unit for stitching together the differences between the multiple interest sub-images and the calibration sub-images to determine the remnants.

[0094] As an optional embodiment, the sixth determining unit includes: a first determining subunit, used to determine a first grayscale mean of the calibration sub-image and a second grayscale mean of the interest sub-image; a second determining subunit, used to determine the grayscale difference between the first grayscale mean and the second grayscale mean; a third determining subunit, used to determine the interest sub-image that differs from the calibration sub-image as a difference sub-image when the ratio of the grayscale difference to the first grayscale mean exceeds a preset grayscale threshold; and a fourth determining subunit, used to extract the foreground from the difference sub-image and determine the extracted foreground as a difference.

[0095] As an optional embodiment, the fourth determining subunit includes: a setting subunit, configured to set pixels in the same region of the difference subimage and the calibration subimage to a first preset value, and set pixels in the difference region of the difference subimage and the calibration subimage to a second preset value, based on a preset background model, wherein the preset background model is determined by initializing the calibration subimage according to the running target detection algorithm, and the region in the difference subimage where the pixels are set to the second preset value is the foreground; and an extraction subunit, configured to extract the region in the difference subimage where the pixels are set to the second preset value as the difference.

[0096] Embodiments of the present invention can provide a computer terminal, which can be any computer terminal device in a group of computer terminals. Optionally, in this embodiment, the computer terminal can also be replaced by a mobile terminal or other terminal device.

[0097] Optionally, in this embodiment, the computer terminal may be located in at least one of a plurality of network devices in a computer network.

[0098] In this embodiment, the computer terminal described above can execute the program code for the following steps in the image detection method: acquiring a detection image of the detection space, wherein the detection image indicates that there is no target object in the detection space; determining a temporal image sequence based on the detection image, wherein the temporal image sequence includes: multiple frames of preset detection images of the detection space, the multiple frames of preset detection images are temporally adjacent, and the detection image is located in the middle of the temporal image sequence; determining the detection image as an interest image when the number of frames belonging to the target detection image in the temporal image sequence does not exceed a preset frame number threshold, wherein the target detection image is a preset detection image containing a target object; determining the difference between the interest image and the calibration image as a remnant of the target object, wherein the calibration image is pre-collected under the condition that there is no remnant or target object in the detection space.

[0099] Optionally, Figure 4 This is a structural block diagram of a computer terminal according to an embodiment of the present invention. Figure 4 As shown, the computer terminal 40 may include one or more (only one is shown in the figure) processors 42 and memory 44.

[0100] The memory can be used to store software programs and modules, such as the program instructions / modules corresponding to the image detection method and apparatus in this embodiment of the invention. The processor executes various functional applications and data processing by running the software programs and modules stored in the memory, thereby realizing the aforementioned image detection method. The memory may include high-speed random access memory, and may also include non-volatile memory, such as one or more magnetic storage devices, flash memory, or other non-volatile solid-state memory. In some instances, the memory may further include memory remotely located relative to the processor, and these remote memories can be connected to the terminal 40 via a network. Examples of such networks include, but are not limited to, the Internet, corporate intranets, local area networks, mobile communication networks, and combinations thereof.

[0101] The processor can invoke information and application programs stored in the memory via a transmission device to perform the following steps: acquiring a detection image of the detection space, wherein the detection image indicates that no target object exists in the detection space; determining a temporal image sequence based on the detection image, wherein the temporal image sequence includes: multiple frames of preset detection images of the detection space, the multiple frames of preset detection images being temporally adjacent, and the detection image being located in the middle of the temporal image sequence; determining the detection image as an interest image if the number of frames belonging to the target detection image in the temporal image sequence does not exceed a preset frame number threshold, wherein the target detection image is a preset detection image containing a target object; determining the difference between the interest image and the calibration image as remnants of the target object, wherein the calibration image is pre-collected under the condition that no remnants or target object exist in the detection space.

[0102] Optionally, the processor may also execute program code that performs the following steps: acquiring a preset detection image of the space to be detected at preset time intervals; detecting whether a target object exists in the preset detection image; and determining the preset detection image as the image to be detected if no target object exists in the preset detection image.

[0103] Optionally, the processor may also execute program code for the following steps: acquiring multiple frames of first detection images that are time-series preceding the image to be detected, wherein the multiple frames of first detection images are time-adjacent, and one of the multiple frames of first detection images is a first detection image that is time-adjacent to the image to be detected; acquiring multiple frames of second detection images that are time-series following the image to be detected, wherein the multiple frames of second detection images are time-adjacent, and one of the multiple frames of second detection images is a second detection image that is time-adjacent to the image to be detected; and determining multiple frames of preset detection images in the time-series image sequence based on the multiple frames of first detection images, the image to be detected, and the multiple frames of second detection images.

[0104] Optionally, the processor may also execute program code that performs the following steps: detects whether a target object exists in each frame of a preset detection image in a time-series image sequence; determines the preset detection image containing the target object as the target detection image; counts the number of frames of the target detection images in the time-series image sequence; and determines the image to be detected as an image of interest if the number of frames of the target detection images does not exceed a preset frame number threshold.

[0105] Optionally, the processor may also execute program code that performs the following steps: dividing the interest image into multiple interest sub-images; determining the calibration sub-image corresponding to the interest sub-image in the calibration image, wherein the calibration image includes: multiple pre-divided calibration sub-images; determining the differences between the interest sub-images and the calibration sub-images; stitching together the differences between the multiple interest sub-images and the calibration sub-images to determine the remnants.

[0106] Optionally, the processor may also execute program code for the following steps: determining the first gray-scale mean of the calibration sub-image and the second gray-scale mean of the interest sub-image; determining the gray-scale difference between the first gray-scale mean and the second gray-scale mean; if the ratio of the gray-scale difference to the first gray-scale mean exceeds a preset gray-scale threshold, determining the interest sub-image that differs from the calibration sub-image as the difference sub-image; and performing foreground extraction on the difference sub-image to determine the extracted foreground as the difference.

[0107] Optionally, the processor may also execute program code for the following steps: based on a preset background model, setting the pixels of the same region in the difference sub-image and the calibration sub-image to a first preset value, and setting the pixels of the difference region in the difference sub-image and the calibration sub-image to a second preset value, wherein the preset background model is determined by initializing the calibration sub-image according to the running target detection algorithm, and the region in the difference sub-image where the pixels are set to the second preset value is the foreground; extracting the region in the difference sub-image where the pixels are set to the second preset value as the difference.

[0108] This invention provides an image detection scheme. The scheme involves: acquiring a target image of a detection space, where the target image indicates the absence of a target object within the detection space; determining a temporal image sequence based on the target image, where the temporal image sequence includes multiple frames of preset detection images of the detection space, with these frames being temporally adjacent and the target image located in the middle of the temporal image sequence; determining the target image as an interest image if the number of frames belonging to the target detection image in the temporal image sequence does not exceed a preset frame threshold, where the target detection image is a preset detection image containing a target object; identifying the difference between the interest image and a calibration image as remnants of the target object, where the calibration image is pre-collected under the condition that neither remnants nor the target object exist in the detection space; thereby enabling the detection of remnants of the target object even when the target object does not frequently appear in the detection space, ensuring that the detected remnants are items left behind by the target object, rather than items used by the target object in the detection space, thus achieving the technical effect of detecting false positives for remnants and solving the problem of false positives in existing remnant detection methods.

[0109] Those skilled in the art will understand that Figure 4 The structure shown is for illustrative purposes only. The computer terminal can also be a smartphone (such as an Android phone, an iOS phone, etc.), a tablet computer, a mobile internet device (MID), a PAD, and other terminal devices. Figure 4 This does not limit the structure of the aforementioned electronic device. For example, computer terminal 40 may also include components that are more... Figure 4 The more or fewer components shown (such as network interfaces, display devices, etc.), or having the same Figure 4 The different configurations shown.

[0110] Those skilled in the art will understand that all or part of the steps in the various methods of the above embodiments can be implemented by a program instructing the hardware related to the terminal device. The program can be stored in a non-volatile medium, which may include: flash drive, read-only memory (ROM), random access memory (RAM), disk or optical disk, etc.

[0111] Embodiments of the present invention also provide a non-volatile storage medium. Optionally, in this embodiment, the non-volatile storage medium can be used to store the program code executed by the image detection method provided in the above embodiments.

[0112] Optionally, in this embodiment, the non-volatile storage medium may be located in any computer terminal in a group of computer terminals in a computer network, or in any mobile terminal in a group of mobile terminals.

[0113] Optionally, in this embodiment, the non-volatile storage medium is configured to store program code for performing the following steps: acquiring a detection image of the detection space, wherein the detection image indicates that no target object exists in the detection space; determining a temporal image sequence based on the detection image, wherein the temporal image sequence includes: multiple frames of preset detection images of the detection space, the multiple frames of preset detection images are temporally adjacent, and the detection image is located in the middle of the temporal image sequence; determining the detection image as an interest image when the number of frames belonging to the target detection image in the temporal image sequence does not exceed a preset frame number threshold, wherein the target detection image is a preset detection image containing a target object; determining the difference between the interest image and the calibration image as a remnant of the target object, wherein the calibration image is pre-collected under the condition that no remnant or target object exists in the detection space.

[0114] Optionally, in this embodiment, the non-volatile storage medium is configured to store program code for performing the following steps: acquiring a preset detection image of the space to be detected at preset time intervals; detecting whether a target object exists in the preset detection image; and determining the preset detection image as the image to be detected if no target object exists in the preset detection image.

[0115] Optionally, in this embodiment, the non-volatile storage medium is configured to store program code for performing the following steps: acquiring multiple frames of first detection images that are time-series preceding the image to be detected, wherein the multiple frames of first detection images are time-adjacent, and there is a first detection image that is time-adjacent to the image to be detected among the multiple frames of first detection images; acquiring multiple frames of second detection images that are time-series following the image to be detected, wherein the multiple frames of second detection images are time-adjacent, and there is a second detection image that is time-adjacent to the image to be detected among the multiple frames of second detection images; and determining multiple frames of preset detection images in the time-series image sequence based on the multiple frames of first detection images, the image to be detected, and the multiple frames of second detection images.

[0116] Optionally, in this embodiment, the non-volatile storage medium is configured to store program code for performing the following steps: detecting whether a target object exists in each frame of a preset detection image in a time-series image sequence; determining that the preset detection image containing the target object is a target detection image; counting the number of frames of the target detection images in the time-series image sequence; and determining that the image to be detected is an image of interest if the number of frames of the target detection images does not exceed a preset frame number threshold.

[0117] Optionally, in this embodiment, the non-volatile storage medium is configured to store program code for performing the following steps: dividing an interest image into multiple interest sub-images; determining a calibration sub-image corresponding to the interest sub-image in a calibration image, wherein the calibration image includes: multiple pre-divided calibration sub-images; determining the differences between the interest sub-images and the calibration sub-images; stitching together the differences between the multiple interest sub-images and the calibration sub-images to determine the remnants.

[0118] Optionally, in this embodiment, the non-volatile storage medium is configured to store program code for performing the following steps: determining a first grayscale mean of the calibration sub-image and a second grayscale mean of the interest sub-image; determining the grayscale difference between the first grayscale mean and the second grayscale mean; if the ratio of the grayscale difference to the first grayscale mean exceeds a preset grayscale threshold, determining the interest sub-image that differs from the calibration sub-image as a difference sub-image; and performing foreground extraction on the difference sub-image to determine the extracted foreground as a difference.

[0119] Optionally, in this embodiment, the non-volatile storage medium is configured to store program code for performing the following steps: based on a preset background model, the pixels of the same region in the difference sub-image and the calibration sub-image are set to a first preset value, and the pixels of the difference region in the difference sub-image and the calibration sub-image are set to a second preset value. The preset background model is determined by initializing the calibration sub-image according to the running target detection algorithm, and the region in the difference sub-image where the pixels are set to the second preset value is the foreground. The region in the difference sub-image where the pixels are set to the second preset value is extracted as the difference.

[0120] The sequence numbers of the above embodiments of the present invention are for descriptive purposes only and do not represent the superiority or inferiority of the embodiments.

[0121] In the above embodiments of the present invention, the descriptions of each embodiment have different focuses. For parts not described in detail in a certain embodiment, please refer to the relevant descriptions of other embodiments.

[0122] In the several embodiments provided in this application, it should be understood that the disclosed technical content can be implemented in other ways. The device embodiments described above are merely illustrative; for example, the division of units can be a logical functional division, and in actual implementation, there may be other division methods. For instance, multiple units or components may be combined or integrated into another system, or some features may be ignored or not executed. Furthermore, the displayed or discussed mutual coupling, direct coupling, or communication connection may be through some interfaces; the indirect coupling or communication connection between units or modules may be electrical or other forms.

[0123] The units described as separate components may or may not be physically separate. The components shown as units may or may not be physical units; that is, they may be located in one place or distributed across multiple units. Some or all of the units can be selected to achieve the purpose of this embodiment according to actual needs.

[0124] Furthermore, the functional units in the various embodiments of the present invention can be integrated into one processing unit, or each unit can exist physically separately, or two or more units can be integrated into one unit. The integrated unit can be implemented in hardware or as a software functional unit.

[0125] If the integrated unit is implemented as a software functional unit and sold or used as an independent product, it can be stored in a non-volatile storage medium. Based on this understanding, the technical solution of the present invention, in essence, or the part that contributes to the prior art, or all or part of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a non-volatile storage medium and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute all or part of the steps of the methods described in the various embodiments of the present invention. The aforementioned non-volatile storage medium includes various media capable of storing program code, such as USB flash drives, read-only memory (ROM), random access memory (RAM), portable hard drives, magnetic disks, or optical disks.

[0126] The above description is only a preferred embodiment of the present invention. It should be noted that for those skilled in the art, several improvements and modifications can be made without departing from the principle of the present invention, and these improvements and modifications should also be considered within the scope of protection of the present invention.

Claims

1. An image detection method, characterized in that, include: Obtain the image to be detected in the space to be detected, wherein the image to be detected indicates that there is no target object in the space to be detected; A temporal image sequence is determined based on the image to be detected, wherein the temporal image sequence includes: multiple frames of preset detection images of the space to be detected, the multiple frames of preset detection images are temporally adjacent, and the image to be detected is located in the middle position of the temporal image sequence; If the number of frames belonging to the target detection image in the time-series image sequence does not exceed a preset frame number threshold, the image to be detected is determined to be an image of interest, wherein the target detection image is the preset detection image containing the target object; The difference between the interest image and the calibration image is determined as the remnant of the target object, wherein the calibration image is acquired in advance under the condition that neither the remnant nor the target object exists in the space to be detected; Determining the time-series image sequence based on the image to be detected includes: Acquire multiple frames of first detection images that are time-series preceding the image to be detected, wherein the multiple frames of first detection images are time-adjacent, and there is a first detection image that is time-adjacent to the image to be detected among the multiple frames of first detection images; Acquire multiple frames of second detection images that are temporally adjacent to the image to be detected, wherein the multiple frames of second detection images are temporally adjacent, and there is a second detection image temporally adjacent to the image to be detected in the multiple frames of second detection images; Based on multiple frames of the first detection image, the image to be detected, and multiple frames of the second detection image, multiple frames of the preset detection image in the time-series image sequence are determined.

2. The method according to claim 1, characterized in that, The image to be detected, which is obtained from the space to be detected, includes: Preset detection images of the space to be detected are acquired at preset time intervals; Detect whether the target object exists within the preset detection image; If the target object is not found in the preset detection image, the preset detection image is determined to be the image to be detected.

3. The method according to claim 1, characterized in that, Determining the image to be detected as an image of interest when the number of frames belonging to the target detection image in the time-series image sequence does not exceed a preset frame number threshold includes: Detect whether the target object exists in each frame of the preset detection image in the time-series image sequence; The preset detection image containing the target object is identified as the target detection image; Count the number of frames of the target detection image in the time-series image sequence; If the number of frames in the target detection image does not exceed the preset frame number threshold, the image to be detected is determined to be the image of interest.

4. The method according to claim 1, characterized in that, The differences between the interest image and the calibration image are identified as remnants of the target object, including: The interest image is divided into multiple interest sub-images; In the calibration image, a calibration sub-map corresponding to the interest sub-map is determined, wherein the calibration image includes: a plurality of pre-divided calibration sub-maps; Determine the differences between the interest subgraph and the calibration subgraph; The differences between multiple interest subgraphs and calibration subgraphs are combined to determine the legacy.

5. The method according to claim 4, characterized in that, Determining the differences between the interest subgraph and the calibration subgraph includes: Determine the first gray-scale mean of the calibration sub-image and the second gray-scale mean of the interest sub-image; Determine the grayscale difference between the first grayscale mean and the second grayscale mean; If the ratio of the grayscale difference to the first grayscale mean exceeds a preset grayscale threshold, the interest sub-map that differs from the calibration sub-map is determined to be a difference sub-map. Foreground extraction is performed on the difference sub-image, and the extracted foreground is determined to be the difference.

6. The method according to claim 5, characterized in that, Foreground extraction is performed on the difference sub-image, and the extracted foreground is determined to be the difference, including: Based on a preset background model, pixels in the same region of the difference sub-image and the calibration sub-image are set to a first preset value, and pixels in the difference region of the difference sub-image and the calibration sub-image are set to a second preset value. The preset background model is determined by initializing the calibration sub-image according to the running target detection algorithm, and the region in the difference sub-image where the pixels are set to the second preset value is the foreground. The region in the difference sub-image where the pixel value is the second preset value is extracted as the difference.

7. An image detection device, characterized in that, include: An acquisition module is used to acquire a detection image of a detection space, wherein the detection image indicates that no target object exists in the detection space; The first determining module is used to determine a temporal image sequence based on the image to be detected, wherein the temporal image sequence includes: multiple frames of preset detection images of the space to be detected, the multiple frames of preset detection images are temporally adjacent, and the image to be detected is located in the middle position of the temporal image sequence; The second determining module is used to determine the image to be detected as an interest image when the number of frames belonging to the target detection image in the time-series image sequence does not exceed a preset frame number threshold, wherein the target detection image is the preset detection image containing the target object; The third determining module is used to determine the difference between the interest image and the calibration image as a remnant of the target object, wherein the calibration image is acquired in advance under the condition that neither the remnant nor the target object exists in the space to be detected; The first determining module includes: The first acquisition unit is used to acquire multiple frames of first detection images that are time-series preceding the image to be detected, wherein the multiple frames of first detection images are time-adjacent, and there is a first detection image that is time-adjacent to the image to be detected among the multiple frames of first detection images. The second acquisition unit is used to acquire multiple frames of second detection images that are temporally adjacent to the image to be detected, wherein the multiple frames of second detection images are temporally adjacent, and there is a second detection image temporally adjacent to the image to be detected in the multiple frames of second detection images. The second determining unit is used to determine multiple frames of the preset detection images in the time-series image sequence based on multiple frames of the first detection image, the image to be detected, and multiple frames of the second detection image.

8. A non-volatile storage medium, characterized in that, The non-volatile storage medium is used to store a program, wherein, when the program is running, it controls the device where the non-volatile storage medium is located to execute the image detection method according to any one of claims 1 to 6.

9. An electronic device, characterized in that, include: A memory and a processor, the processor being configured to run a program stored in the processor, wherein the program, when running, performs the image detection method according to any one of claims 1 to 6.

Images