Method and device for identifying category of video logo

Abstract

The present invention relates to a method and device for classifying video logos, the method comprising: obtaining a UGC image; determining whether there is a logo in the UGC image by a logo target detection model, and obtaining a logo detection result; when the logo detection result is that there is a logo in the UGC image, intercepting a logo area image from the UGC image; extracting features of the logo area image by a target logo feature extraction model to obtain target logo features; determining the corresponding logo category according to the target logo features and a pre-constructed logo feature retrieval library. This technical solution can more accurately recall the logo to be identified, effectively improve the accuracy, and flexibly expand new logo categories to be identified.

Description

Technical Field

[0001] The present invention relates to the technical field of video logo recognition, and in particular to a method and device for identifying the category of a video logo. Background Art

[0002] As a large amount of video information flows into people's real life, video logo detection is an effective means of analyzing the source of videos. Through the video logo, it is relatively easy to determine the publisher of the video, and through the logo in the program, the specific program can be located. Through these important semantic information, accurate video search can be provided. In addition, by detecting the logo in the video program, advertising clips can be removed to improve the viewing experience. At the same time, in the field of video security, video logo detection technology can effectively determine the source of the video and provide an automatic monitoring method for filtering the programs of fixed TV stations.

[0003] Video logo recognition generally relies on a multi-classification target detection model to directly detect and recognize the logo. This detection and recognition method has a high false positive rate because the logo categories are relatively similar. When adding a new logo category, a large amount of corresponding category data needs to be collected again, and a new logo target detection model needs to be retrained. This consumes a lot of manpower and training time. Summary of the invention

[0004] In order to overcome the problems existing in the related art, the present invention provides a method and device for identifying the category of video logos, which can more accurately recall the logos to be identified, effectively improve the accuracy, and flexibly expand new logo categories to be identified.

[0005] According to a first aspect of an embodiment of the present invention, a method for identifying a category of a video logo is provided, the method comprising:

[0006] Get UGC images;

[0007] Determine whether there is a station logo in the UGC image by using a station logo target detection model, and obtain a station logo detection result;

[0008] When the station logo detection result indicates that there is a station logo in the UGC image, intercepting a station logo area image from the UGC image;

[0009] Extracting features of the logo area image using a target logo feature extraction model to obtain features of the target logo;

[0010] According to the target logo feature and a pre-built logo feature retrieval library, a corresponding logo category is determined.

[0011] In one embodiment, preferably, the method further comprises:

[0012] Acquire multiple logo image data;

[0013] Label each station logo image data by its logo category to obtain a labeled station logo image training set;

[0014] Preprocessing each annotated station logo image in the annotated station logo image training set to obtain a processed station logo image;

[0015] Metric learning training is performed based on the processed station logo image to obtain the target station logo feature extraction model.

[0016] In one embodiment, preferably, the method further comprises:

[0017] Acquire multiple newly added categories of to-be-recognized station logo images, wherein the newly added categories of to-be-recognized station logo images are different from the station logo categories in the labeled station logo image training set;

[0018] Using the target logo feature extraction model, feature extraction is performed on the multiple newly added categories of logo images to be identified and the multiple logo image data to obtain target logo features;

[0019] The target logo feature is used to construct the logo feature retrieval library.

[0020] In one embodiment, preferably, determining the corresponding station logo category according to the target station logo feature and a pre-built station logo feature search library includes:

[0021] Comparing the target logo feature with each logo feature in the logo feature retrieval library for similarity to obtain a comparison result;

[0022] All comparison results are sorted in descending order according to similarity, and a preset number of similarity comparison results ranked top are extracted therefrom;

[0023] Determine whether there is a target logo category whose cumulative appearance times exceeds a preset number of times among the preset number of similarity comparison results;

[0024] When the target logo category exists, the target logo category is marked and output on the UGC image.

[0025] In one embodiment, preferably, the method further comprises:

[0026] When the logo detection result is that there is no logo in the UGC image, or when there is no target logo category with a cumulative appearance number exceeding a preset number in the preset number of similarity comparison results, the UGC image is determined to be a normal image.

[0027] According to a second aspect of an embodiment of the present invention, a device for identifying a category of a video logo is provided, the device comprising:

[0028] A first acquisition module is used to acquire a UGC image;

[0029] A judgment module, used for judging whether there is a station logo in the UGC image through a station logo target detection model, and obtaining a station logo detection result;

[0030] A capture module, configured to capture a logo area image from the UGC image when the logo detection result indicates that there is a logo in the UGC image;

[0031] A first extraction module, configured to extract features from the logo area image using a target logo feature extraction model to obtain features of the target logo;

[0032] The determination module is used to determine the corresponding station logo category according to the target station logo feature and a pre-built station logo feature retrieval library.

[0033] In one embodiment, preferably, the device further comprises:

[0034] A second acquisition module, used for acquiring a plurality of station logo image data;

[0035] A labeling module is used to label each station logo image data with a station logo category to obtain a labeled station logo image training set;

[0036] A processing module, used for preprocessing each annotated station logo image in the annotated station logo image training set to obtain a processed station logo image;

[0037] The training module is used to perform metric learning training according to the processed station logo image to obtain the target station logo feature extraction model.

[0038] In one embodiment, preferably, the device further comprises:

[0039] A third acquisition module is used to acquire multiple newly added categories of to-be-recognized station logo images, wherein the newly added categories of to-be-recognized station logo images are different from the station logo categories in the labeled station logo image training set;

[0040] A second extraction module, configured to use the target logo feature extraction model to perform feature extraction on the multiple newly added categories of logo images to be identified and the multiple logo image data to obtain target logo features;

[0041] A construction module is used to construct the station logo feature retrieval library using the target station logo feature.

[0042] In one embodiment, preferably, the determination module includes:

[0043] A comparison unit, used for comparing the target logo feature with each logo feature in the logo feature search library to obtain a comparison result;

[0044] A sorting unit, used to sort all the comparison results in descending order according to the similarity, and extract a preset number of similarity comparison results ranked top;

[0045] A determination unit, configured to determine whether there is a target logo category whose cumulative appearance times exceeds a preset number of times in the preset number of similarity comparison results;

[0046] The output unit is used to mark and output the target logo category on the UGC image when the target logo category exists.

[0047] In one embodiment, preferably, the determining module is further used for:

[0048] When the logo detection result is that there is no logo in the UGC image, or when there is no target logo category with a cumulative appearance number exceeding a preset number in the preset number of similarity comparison results, the UGC image is determined to be a normal image.

[0049] According to a third aspect of an embodiment of the present invention, a device for identifying a category of a video logo is provided, the device comprising:

[0050] processor;

[0051] a memory for storing processor-executable instructions;

[0052] Wherein, the processor is configured to:

[0053] Get UGC images;

[0054] Determine whether there is a station logo in the UGC image by using a station logo target detection model, and obtain a station logo detection result;

[0055] When the station logo detection result indicates that there is a station logo in the UGC image, intercepting a station logo area image from the UGC image;

[0056] Extracting features of the logo area image using a target logo feature extraction model to obtain features of the target logo;

[0057] According to the target logo feature and a pre-built logo feature retrieval library, a corresponding logo category is determined.

[0058] According to a fourth aspect of an embodiment of the present invention, there is provided a computer-readable storage medium having computer instructions stored thereon, wherein the instructions, when executed by a processor, implement the steps of the method as described in any one of the embodiments of the second aspect.

[0059] The technical solution provided by the embodiments of the present invention may have the following beneficial effects:

[0060] In an embodiment of the present invention, a new detection and recognition method is proposed in a video logo recognition and detection scenario. By utilizing the retrieval method, the logo to be recognized can be recalled more accurately, effectively improving the accuracy. New logo categories to be recognized can also be flexibly expanded. It is only necessary to input the logo category data to be recognized into the logo feature extraction model to obtain the features and save them in the logo feature retrieval library, which effectively reduces the workload and can quickly realize the recognition of new logos.

[0061] It is to be understood that the foregoing general description and the following detailed description are exemplary and explanatory only and are not restrictive of the invention. BRIEF DESCRIPTION OF THE DRAWINGS

[0062] The accompanying drawings, which are incorporated in and constitute a part of this specification, illustrate embodiments consistent with the invention and, together with the description, serve to explain the principles of the invention.

[0063] Figure 1 The figure is a flow chart of a method for identifying a category of a video logo according to an exemplary embodiment.

[0064] Figure 2 The flowchart of another method for identifying the category of a video logo is shown according to an exemplary embodiment.

[0065] Figure 3 The flowchart of another method for identifying the category of a video logo is shown according to an exemplary embodiment.

[0066] Figure 4 The flowchart of another method for identifying the category of a video logo is shown according to an exemplary embodiment.

[0067] Figure 5 The figure is a specific flow chart of a method for identifying a category of a video logo according to an exemplary embodiment.

[0068] Figure 6 The figure is a flow chart showing a device for identifying a category of a video logo according to an exemplary embodiment.

[0069] Figure 7 The figure is a flow chart showing another device for identifying the category of a video logo according to an exemplary embodiment.

[0070] Figure 8 The figure is a flow chart showing another device for identifying the category of a video logo according to an exemplary embodiment.

[0071] Fig. 9The present invention is a flowchart of a determination module in a device for identifying a category of a video logo according to an exemplary embodiment. DETAILED DESCRIPTION

[0072] Exemplary embodiments will be described in detail herein, examples of which are shown in the accompanying drawings. When the following description refers to the drawings, the same numbers in different drawings represent the same or similar elements unless otherwise indicated. The embodiments described in the following exemplary embodiments do not represent all embodiments consistent with the present invention. Instead, they are merely examples of devices and methods consistent with some aspects of the present invention as detailed in the appended claims.

[0073] Figure 1 The figure is a flow chart of a method for identifying a category of a video logo according to an exemplary embodiment.

[0074] like Figure 1 As shown, according to a first aspect of an embodiment of the present invention, a method for identifying a category of a video logo is provided, the method comprising:

[0075] Step S111, obtaining a UGC image; after obtaining the UGC image, the UGC image may be preprocessed, the image RGB pixel values ​​are normalized by dividing them by 255, the original aspect ratio of the image is maintained, and the image is resized to 512x512.

[0076] Step S112, determining whether there is a station logo in the UGC image by using a station logo target detection model to obtain a station logo detection result; the station logo target detection model may be a pre-trained model.

[0077] Step S113, when the station logo detection result indicates that there is a station logo in the UGC image, intercepting a station logo area image from the UGC image;

[0078] Step S114, extracting features from the logo area image using a target logo feature extraction model to obtain target logo features;

[0079] Step S115, determining the corresponding station logo category according to the target station logo feature and a pre-built station logo feature retrieval library.

[0080] In this embodiment, in the video logo recognition detection scenario, a new detection and recognition method is proposed. By using the retrieval method, the logo to be recognized can be recalled more accurately, effectively improving the accuracy, and new logo categories to be recognized can be flexibly expanded. It is only necessary to input the logo category data to be recognized into the logo feature extraction model to obtain the features and save them in the logo feature retrieval library, which effectively reduces the workload and can quickly realize the recognition of new logos.

[0081] Figure 2 The flowchart of another method for identifying the category of a video logo is shown according to an exemplary embodiment.

[0082] like Figure 2 As shown, in one embodiment, preferably, the method further comprises:

[0083] Step S211, obtaining a plurality of station logo image data;

[0084] Step S212, labeling each station logo image data with a station logo category to obtain a labeled station logo image training set;

[0085] Step S213, preprocessing each annotated station logo image in the annotated station logo image training set to obtain a processed station logo image;

[0086] Step S214, performing metric learning training according to the processed logo image to obtain the target logo feature extraction model.

[0087] In this embodiment, metric learning training can be performed on the annotated logo image to obtain a target logo feature extraction model, and then feature extraction is performed using the target logo feature extraction model.

[0088] Figure 3 The flowchart of another method for identifying the category of a video logo is shown according to an exemplary embodiment.

[0089] like Figure 3 As shown, in one embodiment, preferably, the method further comprises:

[0090] Step S311, obtaining a plurality of newly added categories of to-be-recognized station logo images, wherein the newly added categories of to-be-recognized station logo images are different from the station logo categories in the labeled station logo image training set;

[0091] Step S312, using the target logo feature extraction model to perform feature extraction on the multiple newly added categories of logo images to be identified and the multiple logo image data to obtain target logo features;

[0092] Step S313: construct the station logo feature retrieval library using the target station logo feature.

[0093] Figure 4 The flowchart of another method for identifying the category of a video logo is shown according to an exemplary embodiment.

[0094] like Figure 4 As shown, in one embodiment, preferably, step S115 includes:

[0095] Step S411, performing a similarity comparison between the target logo feature and each logo feature in the logo feature search library to obtain a comparison result;

[0096] Step S412, sorting all the comparison results in descending order according to the similarity, and extracting a preset number of similarity comparison results ranked top;

[0097] Step S413, determining whether there is a target logo category whose cumulative appearance times exceeds a preset number of times in the preset number of similarity comparison results;

[0098] Step S414: When the target logo category exists, mark and output the target logo category on the UGC image.

[0099] In one embodiment, preferably, the method further comprises:

[0100] When the logo detection result is that there is no logo in the UGC image, or when there is no target logo category with a cumulative appearance number exceeding a preset number in the preset number of similarity comparison results, the UGC image is determined to be a normal image.

[0101] The above technical solution of the present invention is described in detail below with a specific embodiment.

[0102] like Figure 5 As shown, the method for identifying the category of a video logo includes:

[0103] 101. Obtain UGC images;

[0104] 102. Perform image preprocessing, divide the image RGB pixel value by 255 for normalization, keep the original aspect ratio of the image and resize the image to 512x 512;

[0105] 200. Station logo target detection system: detect whether there is a station logo in the image;

[0106] 201. The station logo visual detection model determines whether there is a station logo in the image;

[0107] 202. If the station logo is detected, go to step 401; otherwise, go to step 203;

[0108] 203. If no station logo is detected, the output image is a normal image;

[0109] 300. Construct a station logo feature extraction model and a station logo feature retrieval library: train a model that can extract station logo image features and construct a station logo feature base library for subsequent retrieval and comparison;

[0110] 301. Collecting the image data of the station logo to be identified;

[0111] 302. Labeling the collected station logo image data with station logo categories;

[0112] 303. Divide the image RGB pixel value by 255 for normalization, keep the original aspect ratio of the image and resize the image to 256x 256;

[0113] 304. Input the preprocessed image as a training set into a station logo feature extraction model to perform metric learning training;

[0114] 305. After the training is completed, the target logo feature extraction model is saved for use in extracting features of the logo image;

[0115] 306. The collection of station logo images that need to be recognized has no duplication with the training set. The data volume of each category is greater than 10. If there are new categories that need to be recognized later, continuous collection can be carried out;

[0116] 307. Input the data collected in step 306 into the target logo feature extraction model, extract the features and save them for building a logo feature retrieval library;

[0117] 400. Station logo image retrieval system: determining whether the input station logo area image detected by the station logo detection model is a station logo to be recognized or a normal image;

[0118] 401. Output the target frame of the station logo area detected in step 201;

[0119] 402. intercepting the station logo area in the original image according to the station logo area target frame;

[0120] 403. Input the station logo area into a station logo feature extraction model;

[0121] 404. The station logo feature extracted by the station logo feature extraction model;

[0122] 405. Input the station logo feature to be searched into the constructed station logo feature search library to perform similarity search and comparison;

[0123] 406. Sort the search results in descending order according to similarity, and extract the top 10 results;

[0124] 407. Determine whether the same category of station logos appears more than 5 times in total in the top 10 results;

[0125] 408. Summarize the results. If the same category of logo appears more than 5 times in total, then output the category, otherwise output the image as a normal image.

[0126] Figure 6The figure is a flow chart showing a device for identifying a category of a video logo according to an exemplary embodiment.

[0127] like Figure 6 As shown, according to a second aspect of an embodiment of the present invention, a device for identifying a category of a video logo is provided, the device comprising:

[0128] A first acquisition module 61, used to acquire a UGC image;

[0129] A determination module 62, configured to determine whether there is a station logo in the UGC image by using a station logo target detection model, and obtain a station logo detection result;

[0130] A capture module 63, configured to capture a logo area image from the UGC image when the logo detection result indicates that there is a logo in the UGC image;

[0131] A first extraction module 64, configured to extract features from the logo region image using a target logo feature extraction model to obtain target logo features;

[0132] The determination module 65 is used to determine the corresponding station logo category according to the target station logo feature and a pre-built station logo feature retrieval library.

[0133] Figure 7 The figure is a flow chart showing another device for identifying the category of a video logo according to an exemplary embodiment.

[0134] like Figure 7 As shown, in one embodiment, preferably, the device further comprises:

[0135] The second acquisition module 71 is used to acquire a plurality of station logo image data;

[0136] A labeling module 72, used to label each station logo image data with a station logo category to obtain a labeled station logo image training set;

[0137] A processing module 73, configured to pre-process each annotated station logo image in the annotated station logo image training set to obtain a processed station logo image;

[0138] The training module 74 is used to perform metric learning training according to the processed station logo image to obtain the target station logo feature extraction model.

[0139] Figure 8 The figure is a flow chart showing another device for identifying the category of a video logo according to an exemplary embodiment.

[0140] like Figure 8 As shown, in one embodiment, preferably, the device further comprises:

[0141] A third acquisition module 81 is used to acquire multiple newly added categories of to-be-recognized station logo images, wherein the newly added categories of to-be-recognized station logo images are different from the station logo categories in the labeled station logo image training set;

[0142] A second extraction module 82 is used to extract features from the multiple newly added categories of to-be-recognized logo images and the multiple logo image data using the target logo feature extraction model to obtain target logo features;

[0143] The construction module 83 is used to construct the station logo feature retrieval library using the target station logo feature.

[0144] Fig. 9 The present invention is a flowchart of a determination module in a device for identifying a category of a video logo according to an exemplary embodiment.

[0145] like Fig. 9 As shown, in one embodiment, preferably, the determination module 65 includes:

[0146] A comparison unit 91 is used to compare the target logo feature with each logo feature in the logo feature search library to obtain a comparison result;

[0147] A sorting unit 92 is used to sort all the comparison results in descending order according to the similarity, and extract a preset number of similarity comparison results ranked top;

[0148] A determination unit 93, configured to determine whether there is a target logo category whose cumulative appearance times exceeds a preset number of times in the preset number of similarity comparison results;

[0149] The output unit 94 is used to mark and output the target logo category on the UGC image when the target logo category exists.

[0150] In one embodiment, preferably, the determining module 65 is further used for:

[0151] When the logo detection result is that there is no logo in the UGC image, or when there is no target logo category with a cumulative appearance number exceeding a preset number in the preset number of similarity comparison results, the UGC image is determined to be a normal image.

[0152] According to a third aspect of an embodiment of the present invention, a device for identifying a category of a video logo is provided, the device comprising:

[0153] processor;

[0154] a memory for storing processor-executable instructions;

[0155] Wherein, the processor is configured to:

[0156] Get UGC images;

[0157] Determine whether there is a station logo in the UGC image by using a station logo target detection model, and obtain a station logo detection result;

[0158] When the station logo detection result indicates that there is a station logo in the UGC image, intercepting a station logo area image from the UGC image;

[0159] Extracting features of the logo area image using a target logo feature extraction model to obtain features of the target logo;

[0160] According to the target logo feature and a pre-built logo feature retrieval library, a corresponding logo category is determined.

[0161] According to a fourth aspect of an embodiment of the present invention, there is provided a computer-readable storage medium having computer instructions stored thereon, wherein the instructions, when executed by a processor, implement the steps of the method as described in any one of the embodiments of the second aspect.

[0162] It is further understood that in the present invention, "plurality" refers to two or more than two, and other quantifiers are similar. "And / or" describes the association relationship of associated objects, indicating that three relationships may exist. For example, A and / or B can represent: A exists alone, A and B exist at the same time, and B exists alone. The character " / " generally indicates that the associated objects before and after are in an "or" relationship. The singular forms "a", "the" and "the" are also intended to include plural forms, unless the context clearly indicates other meanings.

[0163] It is further understood that the terms "first", "second", etc. are used to describe various information, but such information should not be limited to these terms. These terms are only used to distinguish the same type of information from each other, and do not indicate a specific order or degree of importance. In fact, the expressions "first", "second", etc. can be used interchangeably. For example, without departing from the scope of the present invention, the first information can also be referred to as the second information, and similarly, the second information can also be referred to as the first information.

[0164] It is further understood that, although the operations are described in a specific order in the accompanying drawings in the embodiments of the present invention, it should not be understood as requiring the operations to be performed in the specific order or serial order shown, or requiring the execution of all the operations shown to obtain the desired results. In certain environments, multitasking and parallel processing may be advantageous.

[0165] Those skilled in the art will readily appreciate other embodiments of the present invention after considering the specification and practicing the invention disclosed herein. This application is intended to cover any variations, uses or adaptations of the present invention that follow the general principles of the present invention and include common knowledge or customary techniques in the art that are not disclosed by the present invention. The specification and examples are to be considered exemplary only, and the true scope and spirit of the present invention are indicated by the following claims.

[0166] It should be understood that the present invention is not limited to the exact construction that has been described above and shown in the drawings and that various modifications and changes may be made without departing from the scope thereof. The scope of the present invention is limited only by the appended claims.

Claims

1. A method for identifying the category of a video logo, characterized in that: The method comprises: Step S111, obtaining a UGC image; Step S112, determining whether there is a station logo in the UGC image by using a station logo target detection model, and obtaining a station logo detection result; Step S113, when the station logo detection result indicates that there is a station logo in the UGC image, intercepting a station logo area image from the UGC image; Step S114, extracting features from the logo area image using a target logo feature extraction model to obtain target logo features; Step S115, determining the corresponding logo category according to the target logo feature and a pre-built logo feature search library; The method further comprises: Step S211, obtaining a plurality of station logo image data; Step S212, labeling each station logo image data with a station logo category to obtain a labeled station logo image training set; Step S213, preprocessing each annotated station logo image in the annotated station logo image training set to obtain a processed station logo image; Step S214, performing metric learning training according to the processed logo image to obtain the target logo feature extraction model; The method further comprises: Step S311, obtaining a plurality of newly added categories of to-be-recognized station logo images, wherein the newly added categories of to-be-recognized station logo images are different from the station logo categories in the labeled station logo image training set; Step S312, using the target logo feature extraction model to perform feature extraction on the multiple newly added categories of logo images to be identified and the multiple logo image data to obtain target logo features; Step S313, constructing the station logo feature retrieval library using the target station logo feature; Determining the corresponding station logo category according to the target station logo feature and the pre-built station logo feature retrieval library includes: Step S411, performing a similarity comparison between the target logo feature and each logo feature in the logo feature search library to obtain a comparison result; Step S412, sorting all the comparison results in descending order according to the similarity, and extracting a preset number of similarity comparison results ranked top; Step S413, determining whether there is a target logo category whose cumulative appearance times exceeds a preset number of times in the preset number of similarity comparison results; Step S414: When the target logo category exists, mark and output the target logo category on the UGC image.

2. The method according to claim 1, characterized in that The method further comprises: When the logo detection result is that there is no logo in the UGC image, or when there is no target logo category with a cumulative appearance number exceeding a preset number in the preset number of similarity comparison results, the UGC image is determined to be a normal image.

3. A video logo category recognition device, characterized in that: The device comprises: A first acquisition module is used to acquire a UGC image; A judgment module, used for judging whether there is a station logo in the UGC image through a station logo target detection model, and obtaining a station logo detection result; A capture module, configured to capture a logo area image from the UGC image when the logo detection result indicates that there is a logo in the UGC image; A first extraction module, configured to extract features from the logo area image using a target logo feature extraction model to obtain features of the target logo; A determination module, used to determine the corresponding station logo category according to the target station logo feature and a pre-built station logo feature retrieval library; The device also includes: A second acquisition module, used for acquiring a plurality of station logo image data; A labeling module is used to label each station logo image data with a station logo category to obtain a labeled station logo image training set; A processing module, used for preprocessing each annotated station logo image in the annotated station logo image training set to obtain a processed station logo image; A training module, used for performing metric learning training according to the processed station logo image to obtain the target station logo feature extraction model; The device also includes: A third acquisition module is used to acquire multiple newly added categories of to-be-recognized station logo images, wherein the newly added categories of to-be-recognized station logo images are different from the station logo categories in the labeled station logo image training set; A second extraction module, configured to use the target logo feature extraction model to perform feature extraction on the multiple newly added categories of logo images to be identified and the multiple logo image data to obtain target logo features; A construction module, used to construct the station logo feature retrieval library using the target station logo feature; The determination module comprises: A comparison unit, used for comparing the target logo feature with each logo feature in the logo feature search library to obtain a comparison result; A sorting unit, used to sort all the comparison results in descending order according to the similarity, and extract a preset number of similarity comparison results ranked top; A determination unit, configured to determine whether there is a target logo category whose cumulative appearance times exceeds a preset number of times in the preset number of similarity comparison results; The output unit is used to mark and output the target logo category on the UGC image when the target logo category exists.

Images