Object detection method and device, object detection system, medium, equipment and product

Abstract

The embodiment of the invention provides an object detection method and device, an object detection system, a medium, equipment and a product, and the method comprises the steps: determining a target detection region in a target image, thereby achieving the positioning of a related region. Due to the fact that true and false objects have large differences in the aspects of gray distribution characteristics and image edge characteristics, namely, the gray distribution of the true objects is dispersed, the gray distribution of the fake objects is concentrated, the image edges of the true objects are smooth, and the edges of the fake objects are not smooth enough, the false objects can be identified. And gray distribution features and image edge features of the target detection area are acquired and further input into a model detection model, and finally, a detection result is determined according to model output.

Description

Technical Field

[0001] This specification relates to the field of image processing technology, and in particular to an object detection method and apparatus, object detection system, medium, equipment and product. Background Technology

[0002] Identification information in digital images (such as stamps representing organizations and individuals) may be altered or used with forged or false information. To improve data security, there is an urgent need for a technical solution to detect the authenticity of identification information in images.

[0003] It should be noted that the information disclosed in the background section above is only used to enhance the understanding of the background of this specification, and therefore may include information that does not constitute prior art known to those skilled in the art. Summary of the Invention

[0004] The purpose of this specification is to provide an object detection method and apparatus, object detection system, medium, equipment and product that can effectively detect the authenticity of identification information in digital images, thereby improving data security to at least a certain extent.

[0005] Other features and advantages of this specification will become apparent from the following detailed description, or may be learned in part by practice of this specification.

[0006] According to one aspect of this specification, an object detection method is provided, the method comprising: determining a target detection region in a target image; acquiring grayscale distribution features of the target detection region and acquiring image edge features of the target detection region; and inputting the grayscale distribution features and image edge features of the target detection region into an object detection model, and determining a detection result of an object to be detected in the target image based on the output of the object detection model, wherein the object to be detected corresponds to the target detection region.

[0007] In an exemplary embodiment, based on the foregoing scheme, determining the target detection region in the target image includes: acquiring multiple candidate boxes corresponding to the target image; and determining a target candidate box in the multiple candidate boxes based on the size features of the object to be detected and the pixel features in the multiple candidate boxes, and determining the region corresponding to the target candidate box in the target image as the target detection region.

[0008] In an exemplary embodiment, based on the foregoing scheme, obtaining multiple candidate boxes corresponding to the target image includes: inputting the target image into a pre-trained target detection model, and outputting the target detection model as multiple candidate boxes corresponding to the target image.

[0009] In an exemplary embodiment, based on the foregoing scheme, the aforementioned size features include: envelope size features and / or aspect ratio features; The method of determining the target candidate box from the multiple candidate boxes based on the size features of the object to be detected and the pixel features of the multiple candidate boxes includes: filtering the multiple candidate boxes based on the envelope size features of the object to be detected and the size of each candidate box in the multiple candidate boxes; and / or filtering the multiple candidate boxes based on the aspect ratio features of the object to be detected and the aspect ratio of each candidate box in the multiple candidate boxes; and determining the target candidate box from the filtered candidate boxes based on the pixel features of the filtered candidate boxes. Among them, the aforementioned pixel features include: the number of pixels corresponding to a candidate box, or the area ratio of pixels in a candidate box.

[0010] In an exemplary embodiment, based on the aforementioned scheme, obtaining the grayscale distribution features of the target detection region includes: obtaining the grayscale of each pixel in the pixel set corresponding to the target detection region; and, based on the grayscale of each pixel in the pixel set, counting the number of pixels corresponding to multiple grayscale levels to obtain the grayscale distribution features of the target detection region.

[0011] In an exemplary embodiment, based on the aforementioned scheme, the acquisition of the image edge features of the target detection region includes: performing edge detection on the target detection region to obtain an edge information map corresponding to the target detection region; segmenting the pixels in the edge information map according to a preset threshold and obtaining pixels with values ​​greater than the preset threshold; and statistically analyzing the grayscale distribution of pixels with values ​​greater than the preset threshold to obtain the image edge features of the target detection region.

[0012] In an exemplary embodiment, based on the foregoing scheme, the above-mentioned edge detection of the target detection region includes: performing edge detection on the target detection region by calculating the Laplacian operator of the target detection region.

[0013] In an exemplary embodiment, based on the foregoing scheme, before inputting the grayscale distribution features and image edge features of the target detection region into the object detection model, the method further includes: determining at least one sample object, and determining at least one forged object corresponding to each sample object; determining positive samples based on the sample objects, and determining negative samples based on the forged objects, wherein the positive samples contain the grayscale distribution features and image edge features of the sample objects, and the negative samples contain the grayscale distribution features and image edge features of the forged objects corresponding to the sample objects; and training a classification algorithm using the positive samples and the negative samples to obtain the object detection model.

[0014] In an exemplary embodiment, based on the aforementioned scheme, the above-mentioned inputting the gray-scale distribution features of the target detection region and the image edge features of the target detection region into the object detection model, and determining the detection result of the object to be detected in the target image based on the output of the object detection model, includes: inputting the gray-scale distribution features of the target detection region and the image edge features of the target detection region into the object detection model; and determining the classification result output by the object detection model as the detection result of the object to be detected in the target image.

[0015] According to another aspect of this specification, an object detection system is provided, the system comprising: a pre-trained object detection model, a feature acquisition module, and an object detection model.

[0016] The pre-trained target detection model is used to determine the target detection region in the target image; the feature acquisition module is used to acquire the gray-level distribution features of the target detection region and the image edge features of the target detection region; and the object detection model is used to receive the gray-level distribution features and the image edge features of the target detection region, and the output of the object detection model is used to determine the detection result of the object to be detected in the target image, wherein the object to be detected corresponds to the target detection region.

[0017] According to another aspect of this specification, an object detection device is provided, the device comprising: a region determination module, a feature acquisition module, and an object detection module.

[0018] The aforementioned region determination module is used to: determine a target detection region in the target image; the aforementioned feature acquisition module is used to: acquire the grayscale distribution features of the aforementioned target detection region and acquire the image edge features of the aforementioned target detection region; and the aforementioned object detection module is used to: input the grayscale distribution features of the aforementioned target detection region and the image edge features of the aforementioned target detection region into an object detection model, and determine the detection result of the object to be detected in the aforementioned target image based on the output of the aforementioned object detection model, wherein the object to be detected corresponds to the aforementioned target detection region.

[0019] According to another aspect of this specification, an electronic device is provided, including a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the computer program to implement the object detection method as described in the above embodiments.

[0020] According to one aspect of this specification, a computer-readable storage medium is provided that stores instructions which, when executed on a computer or processor, cause the computer or processor to perform the object detection method as described in the above embodiments.

[0021] According to another aspect of this specification, a computer program product containing instructions is provided that, when the computer program product is run on a computer or processor, causes the computer or processor to perform the object detection method as described in the above embodiments.

[0022] The object detection method, object detection device, object detection system, computer-readable storage medium, electronic device, and computer program product provided in the embodiments of this specification have the following technical effects: The exemplary embodiments provided in this specification first determine the target detection region in the target image, thereby achieving the localization of the relevant region and improving detection efficiency. Since there are significant differences between genuine and counterfeit objects in terms of grayscale distribution features and image edge features—that is, the grayscale distribution of genuine objects is more dispersed while that of counterfeit objects is more concentrated, and the image edges of genuine objects are smoother while those of counterfeit objects are less smooth—this solution uses the grayscale distribution features and image edge features of the aforementioned target detection region for object detection. This approach is highly scene-specific and can improve detection accuracy. Furthermore, the grayscale distribution features and image edge features of the aforementioned target detection region are input into a detection model, and the detection result is determined based on the model output. Therefore, the solutions provided in the embodiments of this specification can effectively detect the authenticity of identification information in digital images, thereby improving data security.

[0023] It should be understood that the above general description and the following detailed description are exemplary and explanatory only, and are not intended to limit this specification. Attached Figure Description

[0024] The accompanying drawings, which are incorporated in and form part of this specification, illustrate embodiments consistent with this specification and, together with the description, serve to explain the principles of this specification. It is obvious that the drawings described below are merely some embodiments of this specification, and those skilled in the art can obtain other drawings based on these drawings without any inventive effort.

[0025] Figure 1 This is a schematic flowchart of an object detection method provided in one embodiment of this specification.

[0026] Figure 2 This is a flowchart illustrating an embodiment of the object detection system provided in this specification for implementing an object detection method.

[0027] Figure 3 This is a flowchart illustrating a method for determining a target detection region according to an embodiment of this specification.

[0028] Figure 4 This is a schematic diagram of the candidate bounding box corresponding to the target image provided in one embodiment of this specification.

[0029] Figure 5a This is a grayscale distribution histogram that reflects the grayscale distribution characteristics of a real object, provided as an embodiment of this specification.

[0030] Figure 5b This is a grayscale distribution histogram that reflects the grayscale distribution characteristics of a counterfeit object, provided as an embodiment of this specification.

[0031] Figure 6 This is a flowchart illustrating a method for extracting grayscale distribution features of a target detection region according to an embodiment of this specification.

[0032] Figure 7 This is a flowchart illustrating a method for extracting image edge features of a target detection region according to an embodiment of this specification.

[0033] Figure 8a This is a grayscale distribution histogram reflecting the edge features of a forged object image, provided as an embodiment of this specification.

[0034] Figure 8b This is a grayscale distribution histogram reflecting the edge features of a real object image, provided as an embodiment of this specification.

[0035] Figure 9 This is a flowchart illustrating a training method for an object detection model provided in one embodiment of this specification.

[0036] Figure 10 This is a schematic diagram of the structure of an object detection device provided in one embodiment of this specification.

[0037] Figure 11 This is a schematic diagram of the structure of an object detection device provided in another embodiment of this specification.

[0038] Figure 12 This is a schematic diagram of the structure of the electronic device provided in the embodiments of this specification. Detailed Implementation

[0039] To make the objectives, technical solutions, and advantages of this specification clearer, the embodiments of this specification will be described in further detail below with reference to the accompanying drawings.

[0040] In the following description, when referring to the accompanying drawings, the same numbers in different drawings denote the same or similar elements unless otherwise indicated. The embodiments described in the following exemplary embodiments do not represent all embodiments consistent with this specification. Rather, they are merely examples of apparatuses and methods consistent with some aspects of this specification as detailed in the appended claims.

[0041] Example embodiments will now be described more fully with reference to the accompanying drawings. However, example embodiments can be implemented in many forms and should not be construed as limited to the examples set forth herein; rather, these embodiments are provided to make this specification more comprehensive and complete, and to fully convey the concept of example embodiments to those skilled in the art. The described features, structures, or characteristics can be combined in any suitable manner in one or more embodiments. In the following description, numerous specific details are provided to give a full understanding of the embodiments described herein. However, those skilled in the art will recognize that the technical solutions described herein may be practiced with one or more of the specific details omitted, or other methods, components, apparatus, steps, etc., may be employed. In other instances, well-known technical solutions are not shown or described in detail to avoid obscuring various aspects of this specification.

[0042] Furthermore, the accompanying drawings are merely illustrative diagrams of this specification and are not necessarily drawn to scale. The same reference numerals in the drawings denote the same or similar parts, and therefore repeated descriptions of them will be omitted. Some block diagrams shown in the drawings are functional entities and do not necessarily correspond to physically or logically independent entities. These functional entities may be implemented in software, in one or more hardware modules or integrated circuits, or in different network and / or processor devices and / or microcontroller devices.

[0043] For online documents that require customers to upload relevant identification information (such as the seal of a company, school, hospital, or other organization or individual used in ink when signing documents), the detection scheme provided in the embodiments of this specification can be used to identify the authenticity of the identification information in the digital image, thereby ensuring the security of the document.

[0044] Related technologies provide the Mantra-net model for detecting forgery information in digital images. Specifically, it includes a feature extraction network and an anomaly detection network. The feature extraction network acquires image features, such as high-frequency features and structural features. The anomaly detection network detects whether an anomaly exists in the image based on the consistency of image features and the consistency of high-frequency features.

[0045] However, the Mantra-net model is a general algorithm for detecting counterfeit information. It has low specificity for detecting the identification information (such as seals) involved in the embodiments of this specification, resulting in low detection accuracy.

[0046] This specification provides an object detection method, object detection device, object detection system, computer-readable storage medium, electronic device, and computer program product, which can effectively detect the authenticity of identification information in digital images, at least to a certain extent improving data security. Specifically, the following will first... Figure 1 Figure 8 illustrates in detail the embodiments of the object detection method provided in this specification: For example, Figure 1 This is a schematic flowchart illustrating an object detection method provided in one embodiment of this specification. (See reference) Figure 1 The embodiment shown in the figure includes: S110-S130.

[0047] In S110, the target detection region is determined in the target image.

[0048] In an exemplary embodiment, the target image is an image containing the object to be detected. The authenticity of the object to be detected contained in the target image is detected using the scheme provided in the embodiments of this specification. For example, if the target image is a document containing the official seal of Company A, the scheme provided in the embodiments of this specification can effectively detect the authenticity of the official seal of Company A contained in the target image. If it is detected that the official seal of Company A contained in the target image is a forged object (e.g., obtained through graphic design software), rather than a document obtained by applying ink to the official seal of Company A and then stamping it, then it can be determined that the document has no legal validity.

[0049] To improve the detection efficiency of objects in a target image, embodiments of this specification first pre-locate the corresponding objects to be detected in the target image, that is, determine the target detection region in the target image. In an exemplary embodiment, Figure 2 This is a flowchart illustrating an embodiment of an object detection system based on an object detection method provided in this specification. (Reference) Figure 2 The aforementioned image detection system includes: a pre-trained target detection model 210, a feature acquisition module 220, and an object detection model 230.

[0050] In this embodiment, the target image 20 is input into the pre-trained target detection model 210. After processing by the target detection model 210, candidate boxes 22 corresponding to the target image 20 are output. The candidate boxes 22 are then filtered to obtain the target detection region 24. The following will explain... Figure 3 The specific implementation method for determining the target detection region 24 (S110) in the target image 20 will be described in detail.

[0051] In an exemplary embodiment, Figure 3 This is a schematic flowchart illustrating a method for determining a target detection region according to an embodiment of this specification. (See reference...) Figure 3 The embodiment shown in the figure includes: S112, input the target image into the pre-trained target detection model, and output the target detection model as multiple candidate boxes corresponding to the target image.

[0052] For example, the object detection model described above can be a region-based convolutional neural network (R-CNN), Fast R-CNN, or Faster R-CNN. Object detection model 210 uses Faster R-CNN as an example, refer to... Figure 4 After inputting the target image 20 into the pre-trained Faster R-CNN, it outputs multiple candidate boxes corresponding to the target image (including: candidate box 40, candidate box 42, candidate box 44, candidate box 46, candidate box 48 and candidate box 410).

[0053] Furthermore, to improve positioning efficiency and accuracy, in S114 and S114' provided in the embodiments of this specification: the multiple candidate boxes (such as...) are sorted according to the size characteristics of the object to be detected. Figure 4 In the process, candidate boxes 40, 42, 44, 46, 48 and 410 are filtered; further, in S115 provided in the embodiments of this specification: the target candidate box is determined in the above-mentioned candidate boxes according to the pixel features of the multiple candidate boxes, thereby realizing accurate and fast positioning of the target detection area.

[0054] The size characteristics of the object to be detected include: the envelope size characteristics and / or aspect ratio characteristics of the object to be detected.

[0055] Continue to refer to Figure 3 When the size feature of the object to be detected is the envelope size feature of the object to be detected, S114 is executed: the multiple candidate boxes are filtered according to the envelope size feature of the object to be detected and the size of each candidate box in the multiple candidate boxes.

[0056] For example, the envelope size characteristics of the object to be detected can be expressed as: 20×20 ~ 100×100 (unit: millimeters). Reference Figure 4Candidate boxes 40, 42, 44, 46, 48, and 410 whose envelope size is outside the range of 20×20 to 100×100 are eliminated, thereby effectively reducing the number of candidate boxes corresponding to the target image.

[0057] When the size feature of the object to be detected is the aspect ratio feature of the object to be detected, execute S114': filter the multiple candidate boxes according to the aspect ratio feature of the object to be detected and the aspect ratio of each candidate box in the multiple candidate boxes.

[0058] For example, the aspect ratio feature of the object to be detected can be expressed as: between 0.5 and 2. (Reference) Figure 4 Candidate boxes 40, 42, 44, 46, 48, and 410 with aspect ratios outside the range of 0.5 to 2 are filtered out, thereby effectively reducing the number of candidate boxes corresponding to the target image. Of course, if S114 has been executed before S114', the candidate boxes filtered by envelope size feature are then filtered again by aspect ratio feature. Similarly, if S114' has been executed before S114, the candidate boxes filtered by aspect ratio feature are then filtered again by envelope size feature.

[0059] It should be noted that after executing S112, either S114 or S114' can be executed, or S114 can be executed first and then S114', or S114' can be executed first and then S114. This specification does not limit this.

[0060] Continue to refer to Figure 3 In S116, the target candidate box is determined from the filtered candidate boxes based on the pixel features of the filtered candidate boxes.

[0061] For example, refer to Figure 4 If, after the filtering operations of S114 and / or S114', candidate boxes 48 and 410 remain, in this embodiment, the target candidate box is further determined from candidate boxes 48 and 410 based on the pixel features of the candidate boxes.

[0062] The aforementioned pixel features include: the number of pixels corresponding to a candidate box, or the area ratio of pixels in a candidate box. To improve the localization accuracy of the object to be detected, in this embodiment, candidate boxes 410 with a larger area ratio of pixels in the candidate boxes are selected as the aforementioned target candidate boxes.

[0063] After the target candidate box is determined, the region corresponding to the target candidate box in the target image is determined as the target detection region.

[0064] Figure 3 The embodiment shown provides a scheme for determining the target detection region in a target image. This scheme considers the features of the object to be detected during the determination process, thereby improving localization accuracy. Furthermore, locating the target detection region before object detection effectively improves object detection efficiency. Compared to the Mantra-net model in related technologies, which is only effective for detecting forged objects in larger areas of an image but performs poorly in detecting forged objects in smaller areas, the scheme provided in this specification overcomes this problem by considering the size characteristics of the object to be detected, effectively improving detection accuracy and specificity for the detected object.

[0065] Continue to refer to Figure 1 In S120, the grayscale distribution features of the target detection region and the image edge features of the target detection region are obtained.

[0066] There are significant differences between genuine and counterfeit objects in terms of grayscale distribution features and image edge features. Taking grayscale distribution features as an example, genuine objects exhibit a more dispersed grayscale distribution, while counterfeit objects have a more concentrated grayscale distribution. Therefore, this embodiment uses the grayscale distribution features of the target detection region to perform object detection. Detection schemes based on the significant differences in grayscale distribution features between genuine and counterfeit objects have the advantage of strong scene specificity, thereby improving detection accuracy.

[0067] In an exemplary embodiment, the aforementioned grayscale distribution features can be visually displayed using a grayscale distribution histogram. The grayscale histogram is a function of the grayscale distribution, essentially a statistical representation of the grayscale distribution in an image. It calculates the frequency of occurrence of all pixels in a digital image according to their grayscale values. As a function of grayscale levels, the grayscale histogram represents the number of pixels in the image with a specific grayscale level, reflecting the frequency of that particular grayscale value in the image.

[0068] For example, Figure 5a and Figure 5b Gray-level distribution histograms reflecting the gray-level distribution characteristics of real and fake objects are provided for the embodiments of this specification. (Reference) Figure 5a and Figure 5b In a grayscale histogram, the horizontal axis represents the grayscale level, and the vertical axis represents the frequency of occurrence of that grayscale level. (Compare...) Figure 5a and Figure 5bAs can be seen, the pixels in the grayscale histogram of a real object are distributed relatively evenly across multiple grayscale levels, while the pixels in the grayscale histogram of a fake object are concentrated on a smaller number of grayscale levels. In other words, compared to a real object, its distribution across multiple grayscale levels is less even.

[0069] For example, Figure 6 This is a schematic flowchart illustrating a method for extracting grayscale distribution features of a target detection region according to an embodiment of this specification. (See reference...) Figure 6 The embodiment shown in the figure includes: S122, obtain the gray level of each pixel in the pixel set corresponding to the target detection area; and S124, based on the gray level of each pixel in the pixel set, count the number of pixels corresponding to multiple gray levels respectively, and obtain the gray level distribution characteristics of the target detection area.

[0070] Based on the foregoing embodiments, it can be illustrated that grayscale distribution features can serve as an important characteristic for distinguishing the authenticity of an object. Therefore, in order to identify the aforementioned target image (such as...) Figure 2 The authenticity of the target object (target object 20) is determined. In this embodiment, the gray level of each pixel in the target detection area is obtained, and further, the number of pixels corresponding to multiple gray levels is counted to obtain the gray level distribution characteristics of the target detection area.

[0071] For example, the grayscale distribution features of the target detection area obtained above will be input into the object detection model in the scheme provided in the embodiments of this specification to predict the authenticity of the object to be detected.

[0072] In exemplary embodiments, besides the aforementioned grayscale distribution features, there are significant differences in image edge features between genuine and counterfeit objects. The image edges of genuine objects are smoother, while the edges of counterfeit objects are less smooth (in this specification, features reflecting image edge information are referred to as image edge features). This specification embodiment uses the image edge features and grayscale distribution features of the target detection area to perform object detection. Therefore, the detection scheme based on the significant differences between genuine and counterfeit objects has the advantage of strong scene specificity, thereby further improving detection accuracy.

[0073] In an exemplary embodiment, Figure 7 This is a schematic flowchart illustrating a method for extracting image edge features of a target detection region according to an embodiment of this specification. (Reference) Figure 7 The embodiment shown in the figure includes: S122'-S126'.

[0074] In S122', edge detection is performed on the above target detection area to obtain the edge information map corresponding to the above target detection area.

[0075] For example, edge detection of the target detection region can be performed by calculating the Laplacian operator of the target detection region. Edge detection can also be performed using the Sobel operator, Roberts cross gradient operator, or Prewitt operator.

[0076] In S124', the pixels in the edge information map are segmented according to a preset threshold, and pixels larger than the preset threshold are obtained. In S126', the grayscale distribution of pixels larger than the preset threshold is statistically analyzed to obtain the image edge features of the target detection region.

[0077] After edge detection is performed on the target detection area, an image containing the image edge can be generated. In this embodiment, in order to further reduce noise interference, a threshold (such as 150) is set to segment the image containing the image edge and retain pixels with gray values ​​greater than the preset threshold.

[0078] Furthermore, the pixels retained after the aforementioned preset threshold segmentation are statistically analyzed for their distribution within the grayscale value range from the preset threshold to 255. For example, the same... Figure 5a or Figure 5b As shown, a grayscale distribution histogram of the retained pixels is obtained, thereby obtaining features that reflect the image edge information in the above-mentioned target detection area, namely the above-mentioned image edge features.

[0079] For example, Figure 8a and Figure 8b The embodiments of this specification provide grayscale distribution histograms reflecting the edge features of images of forged and real objects. For the image edge features of the forged object, compared to the real image, the number of pixels distributed within the grayscale value range of a preset threshold to 255 is greater, and the distribution across each grayscale level is more uniform (e.g., ...). Figure 8a As shown); conversely, for the image edge features of a real object, compared to a fake image, the number of pixels distributed within the grayscale value preset threshold to 255 is relatively small (sparse), resulting in uneven distribution across different grayscale levels (e.g. Figure 8b (As shown).

[0080] For example, the grayscale distribution features and image edge features of the target detection area obtained above will be input into the object detection model in the scheme provided in the embodiments of this specification to predict the authenticity of the object to be detected.

[0081] In an exemplary embodiment, as previously described, reference is made to... Figure 2If the object detection model 230 is used to preset the authenticity of the object to be detected in the target image, then the solution provided in this embodiment of the specification also includes: training the object detection model 230.

[0082] For example, Figure 9 This is a schematic flowchart illustrating the training method for an object detection model provided in one embodiment of this specification. (See reference) Figure 9 The embodiment shown in the figure includes: S910-S930.

[0083] In S910, at least one sample object is determined, and at least one counterfeit object corresponding to each of the above sample objects is determined.

[0084] The aforementioned sample objects are genuine (such as seals representing their organizations and individuals stamped with ink, including photocopies of the stamps), while the aforementioned counterfeit objects are stamps created using image processing software (such as Photoshop). For example, one of the sample objects is "Dongying xxxx Co., Ltd." (e.g., Figure 4 As shown), further, the chapter is processed using image processing software to obtain various deformed forged objects.

[0085] In S920, a positive sample is determined based on the aforementioned sample object, wherein the positive sample includes the grayscale distribution features and image edge features of the sample object; and in S920', a negative sample is determined based on the aforementioned forged object, wherein the negative sample includes the grayscale distribution features and image edge features of the forged object corresponding to the aforementioned sample object.

[0086] For example, for the above sample object, after as follows Figure 6 The corresponding implementation can determine the grayscale distribution characteristics of the sample object, after such... Figure 7 The corresponding implementation can determine the image edge features of the sample object. Similarly, for the aforementioned forged object, after... Figure 6 The corresponding implementation can determine the grayscale distribution characteristics of the forged object, after such... Figure 7 The corresponding implementation can determine the image edge features of the forged object.

[0087] In S930, the classification algorithm is trained using the positive and negative samples mentioned above to obtain the object detection model.

[0088] Since this technical solution is used to pre-determine the authenticity of the object to be detected, a binary classification model is trained using the above-mentioned positive and negative samples, and the model that obtains the pre-defined evaluation index after training is determined as the above-mentioned object detection model.

[0089] The object detection model provided in the embodiments of this specification is trained based on the gray-scale distribution features of positive and negative samples and image edge features, thereby making the prediction results interpretable.

[0090] Continue to refer to Figure 1 In S130, the grayscale distribution features of the target detection region and the image edge features of the target detection region are input into the object detection model, and the detection result of the target object in the target image is determined based on the output of the object detection model.

[0091] In an exemplary embodiment, for those that have undergone Figure 9 The object detection model trained in the illustrated embodiment takes the grayscale distribution features and image edge features of the target detection region as input, performs internal prediction, and outputs a classification result (specifically, a real object or a fake object), thereby determining the detection result of the object to be detected in the target image.

[0092] In the exemplary embodiments provided in this specification, the target detection region is first determined in the target image, thereby achieving the localization of the relevant region and improving detection efficiency. Then, candidate boxes are filtered based on the size characteristics of the object to be detected, effectively avoiding the problem of small regions being unable to be traversed and detected. Furthermore, since there are significant differences between real and fake objects in terms of grayscale distribution features and image edge features—that is, the grayscale distribution of real objects is more dispersed while that of fake objects is more concentrated, and the image edges of real objects are smoother while those of fake objects are less smooth—this solution uses the aforementioned grayscale distribution features and image edge features of the target detection region for object detection. This approach is highly scene-specific and can improve detection accuracy.

[0093] Furthermore, the grayscale distribution features and image edge features of the target detection region are input into the detection model, and the detection result is determined based on the model output. This model is trained based on the grayscale distribution features and image edge features of positive and negative samples, which enables the prediction results to have strong interpretability.

[0094] As can be seen, the solution provided in the embodiments of this specification can effectively detect the authenticity of the identification information in digital images, thereby improving data security.

[0095] It should be noted that the above figures are merely illustrative of the processes included in the methods according to exemplary embodiments of this specification, and are not intended to be limiting. It is readily understood that the processes shown in the above figures do not indicate or limit the temporal order of these processes. Furthermore, it is readily understood that these processes may, for example, be executed synchronously or asynchronously in multiple modules.

[0096] The following are embodiments of the apparatus described in this specification, which can be used to execute the embodiments of the methods described in this specification. For details not disclosed in the apparatus embodiments of this specification, please refer to the embodiments of the methods described in this specification.

[0097] in, Figure 10 A schematic diagram of an object detection apparatus to which one embodiment of this specification can be applied is shown. Please refer to... Figure 10 The object detection device shown in the figure can be implemented as all or part of an electronic device through software, hardware, or a combination of both. It can also be integrated as an independent module on a server or as an independent module in an electronic device.

[0098] The object detection device 1000 described in the embodiments of this specification includes: a region determination module 1010, a feature acquisition module 1020, and an object detection module 1030.

[0099] The region determination module 1010 is used to: determine the target detection region in the target image; the feature acquisition module 1020 is used to: acquire the gray-level distribution features of the target detection region and the image edge features of the target detection region; and the object detection module 1030 is used to: input the gray-level distribution features and the image edge features of the target detection region into the object detection model, and determine the detection result of the object to be detected in the target image according to the output of the object detection model, wherein the object to be detected corresponds to the target detection region.

[0100] In an exemplary embodiment, Figure 10 A schematic diagram illustrating the structure of an object detection apparatus according to another exemplary embodiment of this specification is provided. Please refer to... Figure 10 : In an exemplary embodiment, based on the aforementioned scheme, the region determination module 1010 includes: a candidate box determination unit 10101 and a candidate box filtering unit 10102.

[0101] The candidate box determination unit 10101 is used to: acquire multiple candidate boxes corresponding to the target image; the candidate box filtering unit 10102 is used to: determine the target candidate box in the multiple candidate boxes according to the size features of the object to be detected and the pixel features in the multiple candidate boxes, and determine the region corresponding to the target candidate box in the target image as the target detection region.

[0102] In an exemplary embodiment, based on the aforementioned scheme, the candidate box determination unit 10101 is specifically used to: input the target image into a pre-trained target detection model, and output the target detection model as multiple candidate boxes corresponding to the target image.

[0103] In an exemplary embodiment, based on the foregoing scheme, the aforementioned size features include: envelope size features and / or aspect ratio features; The candidate box filtering unit 10102 is specifically used for: filtering the plurality of candidate boxes according to the envelope size features of the object to be detected and the size of each candidate box in the plurality of candidate boxes; and / or filtering the plurality of candidate boxes according to the aspect ratio features of the object to be detected and the aspect ratio of each candidate box in the plurality of candidate boxes; and determining the target candidate box from the filtered candidate boxes according to the pixel features of the filtered candidate boxes. Among them, the aforementioned pixel features include: the number of pixels corresponding to a candidate box, or the area ratio of pixels in a candidate box.

[0104] In an exemplary embodiment, based on the foregoing scheme, the feature acquisition module 1020 includes: a grayscale distribution feature acquisition unit 10201.

[0105] The grayscale distribution feature acquisition unit 10201 is used to: acquire the grayscale of each pixel in the pixel set corresponding to the target detection area; and, based on the grayscale of each pixel in the pixel set, count the number of pixels corresponding to multiple grayscale levels to obtain the grayscale distribution features of the target detection area.

[0106] In an exemplary embodiment, based on the foregoing scheme, the feature acquisition module 1020 further includes an image edge feature acquisition unit 10202.

[0107] The image edge feature acquisition unit 10202 is configured to: perform edge detection on the target detection region to obtain an edge information map corresponding to the target detection region; segment the pixels in the edge information map according to a preset threshold and obtain pixels with values ​​greater than the preset threshold; and statistically analyze the grayscale distribution of pixels with values ​​greater than the preset threshold to obtain the image edge features of the target detection region.

[0108] In an exemplary embodiment, based on the aforementioned scheme, the image edge feature acquisition unit 10202 is specifically used to: perform edge detection on the target detection region by calculating the Laplacian operator of the target detection region.

[0109] In an exemplary embodiment, based on the foregoing scheme, the object detection device 1000 further includes a model training module 1040.

[0110] The model training module 1040 is configured to: determine at least one sample object and at least one forged object corresponding to each sample object before inputting the grayscale distribution features and image edge features of the target detection region into the object detection model; determine positive samples based on the sample objects and negative samples based on the forged objects, wherein the positive samples contain the grayscale distribution features and image edge features of the sample objects, and the negative samples contain the grayscale distribution features and image edge features of the forged objects corresponding to the sample objects; and train a classification algorithm using the positive samples and the negative samples to obtain the object detection model.

[0111] In an exemplary embodiment, based on the aforementioned scheme, the object detection module 1030 specifically uses the following methods: inputting the grayscale distribution features of the target detection region and the image edge features of the target detection region into the object detection model; and determining the classification result output by the object detection model as the detection result of the object to be detected in the target image.

[0112] It should be noted that the object detection device provided in the above embodiments is only illustrated by the division of the above functional modules when executing the object detection method. In actual applications, the above functions can be assigned to different functional modules as needed, that is, the internal structure of the device can be divided into different functional modules to complete all or part of the functions described above.

[0113] Furthermore, the object detection device and object detection method embodiments provided in the above embodiments belong to the same concept. Therefore, for details not disclosed in the device embodiments of this specification, please refer to the object detection method embodiments described above, which will not be repeated here.

[0114] The example numbers in this specification are for descriptive purposes only and do not represent the superiority or inferiority of the examples.

[0115] This specification also provides an electronic device, including a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the program to implement the steps of any of the methods described above.

[0116] Figure 12 This schematic diagram illustrates the structure of an electronic device according to an exemplary embodiment of this specification. Please refer to... Figure 12 As shown, the electronic device 1200 includes a processor 1201 and a memory 1202.

[0117] In this embodiment, processor 1201 is the control center of the computer system and can be a processor of a physical machine or a processor of a virtual machine. Processor 1201 may include one or more processing cores, such as a 4-core processor or an 8-core processor. Processor 1201 may be implemented using at least one hardware form selected from Digital Signal Processing (DSP), Field-Programmable Gate Array (FPGA), and Programmable Logic Array (PLA). Processor 1201 may also include a main processor and a coprocessor; the main processor is used to process data in the wake-up state, and the coprocessor is a low-power processor used to process data in the standby state.

[0118] In the embodiments described in this specification, the processor 1201 is specifically used for: A target detection region is determined in the target image; the gray-level distribution features and image edge features of the target detection region are obtained; the gray-level distribution features and image edge features of the target detection region are input into an object detection model, and the detection result of the object to be detected in the target image is determined according to the output of the object detection model, wherein the object to be detected corresponds to the target detection region.

[0119] Furthermore, the above-mentioned determination of the target detection region in the target image includes: acquiring multiple candidate boxes corresponding to the target image; and determining a target candidate box in the multiple candidate boxes based on the size features of the object to be detected and the pixel features in the multiple candidate boxes, and determining the region corresponding to the target candidate box in the target image as the target detection region.

[0120] Furthermore, based on the aforementioned scheme, the above-mentioned acquisition of multiple candidate boxes corresponding to the target image includes: inputting the target image into a pre-trained target detection model, and outputting the target detection model as multiple candidate boxes corresponding to the target image.

[0121] Furthermore, the aforementioned dimensional features include: envelope dimensional features and / or aspect ratio features; The method of determining the target candidate box from the multiple candidate boxes based on the size features of the object to be detected and the pixel features of the multiple candidate boxes includes: filtering the multiple candidate boxes based on the envelope size features of the object to be detected and the size of each candidate box in the multiple candidate boxes; and / or filtering the multiple candidate boxes based on the aspect ratio features of the object to be detected and the aspect ratio of each candidate box in the multiple candidate boxes; and determining the target candidate box from the filtered candidate boxes based on the pixel features of the filtered candidate boxes. Among them, the aforementioned pixel features include: the number of pixels corresponding to a candidate box, or the area ratio of pixels in a candidate box.

[0122] Furthermore, the above-mentioned acquisition of the grayscale distribution features of the target detection region includes: acquiring the grayscale of each pixel in the pixel set corresponding to the target detection region; and, based on the grayscale of each pixel in the pixel set, counting the number of pixels corresponding to multiple grayscale levels respectively to obtain the grayscale distribution features of the target detection region.

[0123] Furthermore, the above-mentioned acquisition of image edge features of the target detection region includes: performing edge detection on the target detection region to obtain an edge information map corresponding to the target detection region; segmenting the pixels in the edge information map according to a preset threshold and obtaining pixels with values ​​greater than the preset threshold; and statistically analyzing the grayscale distribution of pixels with values ​​greater than the preset threshold to obtain image edge features of the target detection region.

[0124] Furthermore, the aforementioned edge detection of the target detection region includes: performing edge detection on the target detection region by calculating the Laplacian operator of the target detection region.

[0125] Furthermore, the aforementioned processor 1201 is specifically used for: Before inputting the grayscale distribution features and image edge features of the target detection region into the object detection model, at least one sample object is determined, and at least one forged object corresponding to each sample object is determined; positive samples are determined based on the sample objects, and negative samples are determined based on the forged objects, wherein the positive samples contain the grayscale distribution features and image edge features of the sample objects, and the negative samples contain the grayscale distribution features and image edge features of the forged objects corresponding to the sample objects; and a classification algorithm is trained using the positive samples and the negative samples to obtain the object detection model.

[0126] Furthermore, the above-mentioned inputting the gray-level distribution features and image edge features of the target detection region into the object detection model, and determining the detection result of the object to be detected in the target image based on the output of the object detection model, includes: inputting the gray-level distribution features and image edge features of the target detection region into the object detection model; and determining the classification result output by the object detection model as the detection result of the object to be detected in the target image.

[0127] Memory 1202 may include one or more computer-readable storage media, which may be non-transitory. Memory 1202 may also include high-speed random access memory and non-volatile memory, such as one or more disk storage devices or flash memory devices. In some embodiments of this specification, the non-transitory computer-readable storage media in memory 1202 is used to store at least one instruction for execution by processor 1201 to implement the methods in the embodiments of this specification.

[0128] In some embodiments, the electronic device 1200 further includes a peripheral device interface 1203 and at least one peripheral device. The processor 1201, memory 1202, and peripheral device interface 1203 can be connected via a bus or signal line. Each peripheral device can be connected to the peripheral device interface 1203 via a bus, signal line, or circuit board. Specifically, the peripheral device includes at least one of a display screen 1204, a camera 1205, and an audio circuit 1206.

[0129] Peripheral interface 1203 can be used to connect at least one input / output (I / O) related peripheral device to processor 1201 and memory 1202. In some embodiments of this specification, processor 1201, memory 1202, and peripheral interface 1203 are integrated on the same chip or circuit board; in some other embodiments of this specification, any one or two of processor 1201, memory 1202, and peripheral interface 1203 can be implemented on separate chips or circuit boards. This specification does not specifically limit the embodiments in this regard.

[0130] Display screen 1204 is used to display a user interface (UI). The UI may include graphics, text, icons, videos, and any combination thereof. When display screen 1204 is a touch display screen, it also has the ability to collect touch signals on or above its surface. These touch signals can be input as control signals to processor 1201 for processing. In this case, display screen 1204 can also be used to provide virtual buttons and / or a virtual keyboard, also known as soft buttons and / or a soft keyboard. In some embodiments of this specification, there may be one display screen 1204, which serves as the front panel of electronic device 1200; in other embodiments, there may be at least two display screens 1204, respectively disposed on different surfaces of electronic device 1200 or in a folded design; in still other embodiments, display screen 1204 may be a flexible display screen, disposed on a curved or folded surface of electronic device 1200. Furthermore, display screen 1204 may also be configured as a non-rectangular irregular shape, i.e., a non-rectangular screen. The display screen 1204 can be made of materials such as liquid crystal display (LCD) and organic light-emitting diode (OLED).

[0131] Camera 1205 is used to capture images or videos. Optionally, camera 1205 includes a front-facing camera and a rear-facing camera. Typically, the front-facing camera is located on the front panel of the electronic device, and the rear-facing camera is located on the back of the electronic device. In some embodiments, there are at least two rear-facing cameras, which are any one of a main camera, a depth-sensing camera, a wide-angle camera, and a telephoto camera, to achieve background blurring by fusion of the main camera and the depth-sensing camera, panoramic shooting by fusion of the main camera and the wide-angle camera, virtual reality (VR) shooting, or other fusion shooting functions. In some embodiments of this specification, camera 1205 may also include a flash. The flash can be a single-color temperature flash or a dual-color temperature flash. A dual-color temperature flash refers to a combination of a warm light flash and a cool light flash, which can be used for light compensation at different color temperatures.

[0132] The audio circuit 1206 may include a microphone and a speaker. The microphone is used to collect sound waves from the user and the environment, and convert the sound waves into electrical signals that are input to the processor 1201 for processing. For stereo sound acquisition or noise reduction purposes, there may be multiple microphones, each located in a different part of the electronic device 1200. The microphone may also be an array microphone or an omnidirectional microphone.

[0133] Power supply 1207 is used to supply power to various components in electronic device 1200. Power supply 1207 can be AC ​​power, DC power, a disposable battery, or a rechargeable battery. When power supply 1207 includes a rechargeable battery, the rechargeable battery can be a wired rechargeable battery or a wireless rechargeable battery. A wired rechargeable battery is a battery that is charged via a wired line, and a wireless rechargeable battery is a battery that is charged via a wireless coil. The rechargeable battery can also be used to support fast charging technology.

[0134] The block diagrams of the electronic device shown in the embodiments of this specification do not constitute a limitation on the electronic device 1200. The electronic device 1200 may include more or fewer components than shown, or combine certain components, or use different component arrangements.

[0135] In the description of this specification, it should be understood that the terms "first," "second," etc., are used for descriptive purposes only and should not be construed as indicating or implying relative importance. Those skilled in the art can understand the specific meaning of these terms in this specification based on the specific circumstances. Furthermore, in the description of this specification, unless otherwise stated, "multiple" means two or more. "And / or" describes the relationship between related objects, indicating that three relationships can exist. For example, A and / or B can represent: A alone, A and B simultaneously, or B alone. The character " / " generally indicates that the preceding and following related objects are in an "or" relationship.

[0136] This specification also provides a computer-readable storage medium storing instructions that, when executed on a computer or processor, cause the computer or processor to perform one or more steps in the above embodiments. If the constituent modules of the object detection device described above are implemented as software functional units and sold or used as independent products, they can be stored in the aforementioned computer-readable storage medium.

[0137] In the above embodiments, implementation can be achieved, in whole or in part, through software, hardware, firmware, or any combination thereof. When implemented in software, it can be implemented, in whole or in part, as a computer program product. The computer program product includes one or more computer instructions. When these computer program instructions are loaded and executed on a computer, all or part of the processes or functions described in the embodiments of this specification are generated. The computer can be a general-purpose computer, a special-purpose computer, a computer network, or other programmable device. The computer instructions can be stored in or transmitted through a computer-readable storage medium. The computer instructions can be transmitted from one website, computer, server, or data center to another website, computer, server, or data center via wired (e.g., coaxial cable, fiber optic, Digital Subscriber Line (DSL)) or wireless (e.g., infrared, wireless, microwave, etc.) means. The computer-readable storage medium can be any available medium accessible to a computer or a data storage device such as a server or data center that integrates one or more available media. The aforementioned available media can be magnetic media (e.g., floppy disks, hard disks, magnetic tapes), optical media (e.g., Digital Versatile Discs (DVDs)), or semiconductor media (e.g., Solid State Disks (SSDs)).

[0138] It should be noted that the above description describes specific embodiments of this specification. Other embodiments are within the scope of the appended claims. In some cases, the actions or steps recorded in the claims may be performed in a different order than that shown in the embodiments and still achieve the desired results. Furthermore, the processes depicted in the drawings do not necessarily require the specific or sequential order shown to achieve the desired results. In some embodiments, multitasking and parallel processing are also possible or may be advantageous.

[0139] The above description is merely a specific embodiment of this specification, but the scope of protection of this specification is not limited thereto. Any variations or substitutions that can be easily conceived by those skilled in the art within the scope of the technology disclosed in this specification should be included within the scope of protection of this specification. Therefore, equivalent variations made in accordance with the claims of this specification are still within the scope of this specification.

Claims

1. An object detection method, wherein, The method includes: Identify the target detection region in the target image; The grayscale distribution features of the target detection region and the image edge features of the target detection region are obtained; wherein, the image edge features are obtained through the grayscale distribution of the edge information map corresponding to the target detection region, and the image edge features characterize the smoothness of the image edge of the object to be detected; The grayscale distribution features and image edge features of the target detection region are input into the object detection model, and the detection result of the object to be detected in the target image is determined according to the output of the object detection model. The object to be detected corresponds to the target detection region.

2. The method according to claim 1, wherein, Determining the target detection region in the target image includes: Obtain multiple candidate bounding boxes corresponding to the target image; Based on the size characteristics of the object to be detected and the pixel characteristics of the multiple candidate boxes, a target candidate box is determined from the multiple candidate boxes, and the region corresponding to the target candidate box in the target image is determined as the target detection region.

3. The method according to claim 2, wherein, The process of obtaining multiple candidate boxes corresponding to the target image includes: The target image is input into a pre-trained target detection model, and the output of the target detection model is used as multiple candidate boxes corresponding to the target image.

4. The method according to claim 2, wherein, The dimensional features include: envelope dimensional features and / or aspect ratio features; The step of determining the target candidate box based on the size features of the object to be detected and the pixel features of the multiple candidate boxes includes: The candidate boxes are filtered based on the envelope size features of the object to be detected and the size of each candidate box in the candidate boxes; and / or, the candidate boxes are filtered based on the aspect ratio features of the object to be detected and the aspect ratio of each candidate box in the candidate boxes. The target candidate box is determined from the filtered candidate boxes based on their pixel features. The pixel features include: the number of pixels corresponding to a candidate box, or the area ratio of pixels in a candidate box.

5. The method according to claim 1, wherein, The step of obtaining the grayscale distribution features of the target detection region includes: Obtain the grayscale value of each pixel in the pixel set corresponding to the target monitoring area; Based on the gray level of each pixel in the pixel set, the number of pixels corresponding to multiple gray levels is counted to obtain the gray level distribution characteristics of the target detection area.

6. The method according to claim 1, wherein, The step of obtaining the image edge features of the target detection region includes: By calculating the Laplacian operator of the target detection region, edge detection is performed on the target detection region to obtain the image edge features of the target detection region.

7. The method according to any one of claims 1 to 6, wherein, Before inputting the grayscale distribution features and image edge features of the target detection region into the object detection model, the method further includes: Identify at least one sample object, and identify at least one forged object corresponding to each sample object; Positive samples are determined based on the sample object, and negative samples are determined based on the forged object, wherein the positive sample includes the grayscale distribution features and image edge features of the sample object, and the negative sample includes the grayscale distribution features and image edge features of the forged object corresponding to the sample object; The object detection model is obtained by training the classification algorithm using the positive and negative samples.

8. The method according to claim 7, wherein, The step of inputting the grayscale distribution features and image edge features of the target detection region into the object detection model, and determining the detection result of the object to be detected in the target image based on the output of the object detection model, includes: The grayscale distribution features of the target detection region and the image edge features of the target detection region are input into the object detection model; The classification result output by the object detection model is determined as the detection result of the object to be detected in the target image.

9. An object detection system, wherein, The system includes: A pre-trained object detection model is used to determine the object detection region in the object image; The feature acquisition module is used to acquire the grayscale distribution features of the target detection region and the image edge features of the target detection region; wherein, the image edge features are obtained through the grayscale distribution of the edge information map corresponding to the target detection region, and the image edge features characterize the smoothness of the image edge of the object to be detected; An object detection model is used to receive the grayscale distribution features of the target detection region and the image edge features of the target detection region, and the output of the object detection model is used to determine the detection result of the object to be detected in the target image, wherein the object to be detected corresponds to the target detection region.

10. An object detection device, wherein, The device includes: The region determination module is used to: determine the target detection region in the target image; The feature acquisition module is used to: acquire the grayscale distribution features of the target detection region and acquire the image edge features of the target detection region; wherein, the image edge features are obtained through the grayscale distribution of the edge information map corresponding to the target detection region, and the image edge features characterize the smoothness of the image edge of the object to be detected; The object detection module is used to: input the grayscale distribution features of the target detection region and the image edge features of the target detection region into the object detection model, and determine the detection result of the object to be detected in the target image according to the output of the object detection model, wherein the object to be detected corresponds to the target detection region.

11. A computer-readable storage medium storing instructions that, when executed on a computer or processor, cause the computer or processor to perform the object detection method as described in any one of claims 1 to 8.

12. An electronic device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein, When the processor executes the computer program, it implements the object detection method as described in any one of claims 1 to 8.

13. A computer program product comprising instructions that, when run on a computer or processor, causes the computer or processor to perform the object detection method as described in any one of claims 1 to 8.

Images