Feature library updating method and device, electronic equipment, storage medium and program

Abstract

The embodiment of the invention discloses a feature library updating method and device, electronic equipment, a storage medium and a program, and the method comprises the steps: carrying out the image segmentation processing of a current frame target image, and obtaining the target image features of a current target; determining a target reliability score and a target occlusion degree score of the target image feature; updating a target feature library according to the target reliability score and the target shielding degree score; wherein the target feature library is used for performing image segmentation processing on the current frame target image. According to the embodiment of the invention, the quality of the image features in the feature library can be improved, and the accuracy of video target segmentation is improved.

Description

Technical Field

[0001] The embodiments of the present invention relate to the field of image processing technology, and in particular to a feature library update method, apparatus, electronic device, storage medium and program. Background Technology

[0002] In today's rapidly developing digital media technology, video object segmentation, as a key step in video post-processing, aims to achieve precise separation of foreground objects from background areas through frame-by-frame analysis.

[0003] In existing technologies, to optimize video object segmentation, a method based on building a feature library using neural networks is mainly adopted. Specifically, the neural network comprehensively analyzes historical video frame sequences, real-time input images, and prediction masks generated by the network to extract image features, and stores these feature information in a feature library to provide a reference for the segmentation processing of subsequent frames.

[0004] In the process of realizing this invention, the inventors discovered the following defects in the prior art: when the target is in motion, occlusion and drift are very likely to occur, which will greatly reduce the quality of the image features stored in the feature library, and thus lead to a decrease in the accuracy of video target segmentation results, making it difficult to meet the actual needs of high-precision video editing. Summary of the Invention

[0005] This invention provides a feature library update method, apparatus, electronic device, storage medium, and program that can improve the quality of image features in the feature library, thereby improving the accuracy of video target segmentation.

[0006] According to one aspect of the present invention, a feature library update method is provided, comprising:

[0007] Perform image segmentation processing on the target image of the current frame to obtain the target image features of the current target;

[0008] Determine the target reliability score and target occlusion score of the target image features;

[0009] The target feature library is updated based on the target reliability score and the target occlusion score.

[0010] The target feature library is used to perform image segmentation processing on the target image of the current frame.

[0011] According to another aspect of the present invention, a feature library updating apparatus is provided, comprising:

[0012] The target image feature acquisition module is used to perform image segmentation processing on the target image of the current frame to obtain the target image features of the current target.

[0013] The target image feature scoring and determination module is used to determine the target reliability score and target occlusion score of the target image features.

[0014] The target feature library update module is used to update the target feature library based on the target reliability score and the target occlusion score;

[0015] The target feature library is used to perform image segmentation processing on the target image of the current frame.

[0016] According to another aspect of the present invention, an electronic device is provided, the electronic device comprising:

[0017] At least one processor; and

[0018] A memory communicatively connected to the at least one processor; wherein,

[0019] The memory stores a computer program that can be executed by the at least one processor, the computer program being executed by the at least one processor to enable the at least one processor to perform the feature library update method according to any embodiment of the present invention.

[0020] According to another aspect of the present invention, a computer-readable storage medium is provided, the computer-readable storage medium storing computer instructions for causing a processor to execute and implement the feature library update method according to any embodiment of the present invention.

[0021] According to another aspect of the present invention, a computer program product is also provided, comprising a computer program that, when executed by a processor, implements the feature library update method described in any embodiment of the present invention.

[0022] This invention employs a target feature library to perform image segmentation on the target image of the current frame, thereby obtaining the target image features of the current target. After obtaining the target image features of the current target, a target reliability score and a target occlusion score are determined for the target image features, and the target feature library is updated based on these scores. This scheme, by dynamically updating the target feature library, ensures that the feature library always stores high-quality image features that are highly adapted to the current video segmentation scenario. This solves the problem of poor image feature quality in the feature library in existing technologies, improves the quality of image features in the feature library, and thus improves the accuracy of video target segmentation.

[0023] It should be understood that the description in this section is not intended to identify key or essential features of the embodiments of the present invention, nor is it intended to limit the scope of the invention. Other features of the invention will become readily apparent from the following description. Attached Figure Description

[0024] To more clearly illustrate the technical solutions in the embodiments of the present invention, the accompanying drawings used in the description of the embodiments will be briefly introduced below. Obviously, the accompanying drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0025] Figure 1 This is a flowchart of a feature library update method provided in Embodiment 1 of the present invention;

[0026] Figure 2 This is a flowchart of a feature library update method provided in Embodiment 2 of the present invention;

[0027] Figure 3 This is a flowchart of a specific feature library update method provided in Embodiment 2 of the present invention;

[0028] Figure 4 This is a schematic diagram of a feature library updating device provided in Embodiment 3 of the present invention;

[0029] Figure 5 This is a schematic diagram of the structure of an electronic device provided in Embodiment 4 of the present invention. Detailed Implementation

[0030] To enable those skilled in the art to better understand the present invention, the technical solutions of the present invention will be clearly and completely described below with reference to the accompanying drawings of the embodiments of the present invention. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort should fall within the scope of protection of the present invention.

[0031] It should be noted that the terms "first," "second," and "target," etc., in the specification, claims, and accompanying drawings of this invention are used to distinguish similar objects and are not necessarily used to describe a specific order or sequence. It should be understood that such data can be interchanged where appropriate so that embodiments of the invention described herein can be implemented in orders other than those illustrated or described herein. Furthermore, the terms "comprising" and "having," and any variations thereof, are intended to cover non-exclusive inclusion; for example, a process, method, system, product, or apparatus that comprises a series of steps or units is not necessarily limited to those steps or units explicitly listed, but may include other steps or units not explicitly listed or inherent to such processes, methods, products, or apparatus.

[0032] Example 1

[0033] Figure 1 This is a flowchart of a feature library update method provided in Embodiment 1 of the present invention. This embodiment is applicable to situations where the target feature library is dynamically updated based on the target image of the current frame. This method can be executed by a feature library update device, which can be implemented in software and / or hardware, and is generally integrated into an electronic device. This electronic device can be a terminal device or a server device, as long as it can execute the feature library update method. The present invention does not limit the specific type of electronic device. Correspondingly, as... Figure 1 As shown, the method includes the following operations:

[0034] S110. Perform image segmentation processing on the target image of the current frame to obtain the target image features of the current target;

[0035] The target image of the current frame can be the image to be segmented from the video object. The current target can be a target feature in the target image of the current frame. The target image feature can be the image feature about the current target extracted from the target image of the current frame.

[0036] In this embodiment of the invention, the image to be segmented into a video object can be used as the target image of the current frame. After determining the target image of the current frame, image segmentation processing can be performed on the target image of the current frame to obtain the target image features of the current target in the target image of the current frame.

[0037] S120. Determine the target reliability score and target occlusion score of the target image features.

[0038] The target reliability score is a quantitative assessment of the reliability of the target image features of the current target. The target occlusion score is a quantitative assessment of the degree of occlusion of the target image features of the current target.

[0039] Correspondingly, after performing image segmentation on the target image of the current frame to obtain the target image features of the current target, the target reliability score and the target occlusion score of the target image features of the current target can be calculated.

[0040] In a specific example, a machine learning model from a target feature library can be used to perform image segmentation on the target image in the current frame to obtain the target image features of the current target. Furthermore, after obtaining the target image features of the current target, the machine learning model can output a target reliability score and a target occlusion score for the target image features, so as to dynamically update the image features in the target feature library based on these scores.

[0041] Understandably, a machine learning model can be pre-trained based on target image sample data so that it can be used to perform image segmentation on the target image in the current frame.

[0042] S130. Update the target feature library according to the target reliability score and the target occlusion degree score; wherein, the target feature library is used to perform image segmentation processing on the target image of the current frame.

[0043] The target feature library can be a collection used to store image features that match the current target, as well as to perform image segmentation processing on the target image of the current frame.

[0044] Correspondingly, after determining the target reliability score and target occlusion score of the target image features of the current target, it is possible to determine whether it is necessary to replace the image features in the target feature library with the current target image features based on the target reliability score and target occlusion score of the current target image features.

[0045] In a specific example, in the field of autonomous driving, vehicles need to perform real-time image segmentation of the surrounding environment to identify objects such as pedestrians, other vehicles, traffic lights, and lane lines. Therefore, when performing image segmentation on traffic light video images, the current frame of the traffic light video image can be used as the target image of the current frame, the traffic light can be used as the current target, and the image features related to the current target, such as the brightness features and shape features of the three colors (red, yellow, and green) and the appearance features of different types of traffic lights, can be stored in the target feature library.

[0046] This invention employs a target feature library to perform image segmentation on the target image of the current frame, thereby obtaining the target image features of the current target. After obtaining the target image features of the current target, a target reliability score and a target occlusion score are determined for the target image features, and the target feature library is updated based on these scores. This scheme, by dynamically updating the target feature library, ensures that the feature library always stores high-quality image features that are highly adapted to the current video segmentation scenario. This solves the problem of poor image feature quality in the feature library in existing technologies, improves the quality of image features in the feature library, and thus improves the accuracy of video target segmentation.

[0047] Example 2

[0048] Figure 2 This is a flowchart of a feature library update method provided in Embodiment 2 of the present invention. This embodiment is a specific embodiment based on the above embodiment. In this embodiment, specific optional implementation methods for updating the target feature library according to the target reliability score and the target occlusion degree score are given. Correspondingly, as Figure 2 As shown, the method in this embodiment may include:

[0049] S210. Perform image segmentation processing on the target image of the current frame to obtain the target image features of the current target.

[0050] S220. Determine the target reliability score and target occlusion score of the target image features.

[0051] S230. Determine whether the target reliability score is less than a preset reliability threshold. If yes, execute S240; otherwise, execute S250.

[0052] The preset reliability threshold can be a threshold for the reliability score of the target image features set in advance. This embodiment of the invention does not limit the specific value of the preset reliability threshold.

[0053] S240. Update the target feature library according to the target image features using a preset replacement rule.

[0054] The preset replacement rule can be a pre-set replacement rule for image features in the target feature library.

[0055] Specifically, after obtaining the target reliability score, the target reliability score can be judged. If the target reliability score is less than a preset reliability threshold, the corresponding image features in the target feature library can be replaced using the target image features according to preset replacement rules.

[0056] Figure 3 This is a flowchart of a specific feature library update method provided in Embodiment 2 of the present invention. In a specific example, such as Figure 3 As shown, if the target reliability score is less than the preset reliability threshold, it indicates that there is no tracking target (i.e., the current target) in the current frame image for image segmentation processing. In this case, the image features of the current frame target image can be used as negative sample features, and the corresponding target sample features in the target feature library can be replaced using the image features of the current frame target image according to the preset replacement rule. This reduces the probability of false tracking even when the current target is not present in the current frame target image. It is understandable that the positive sample features in the target feature library all originate from the feature extraction results of images containing the current target.

[0057] S250. Update the target feature library based on the target occlusion score and the target image features.

[0058] Correspondingly, if the target reliability score is determined to be greater than or equal to the preset reliability threshold, it indicates that the current target exists in the current image. Then, the target occlusion score can be further judged, and the target feature library can be updated based on the judgment result of the target occlusion score.

[0059] In an optional embodiment of the present invention, updating the target feature library based on the target occlusion score and the target image features may include: updating the target feature library based on the target image features according to the preset replacement rule when the target occlusion score is determined to be less than or equal to a preset occlusion threshold; and prohibiting updating the target feature library when the target occlusion score is determined to be greater than the preset occlusion threshold.

[0060] The preset occlusion threshold can be a pre-set threshold for the degree of occlusion of the target image features. This embodiment of the invention does not limit the specific value of the preset occlusion threshold.

[0061] In this embodiment of the invention, when it is determined that the current target exists in the target image of the current frame, the target feature library can be updated according to the target occlusion degree score of the target image features of the current target. If the target occlusion degree score is less than or equal to a preset occlusion threshold, it indicates that the current target is not severely occluded in the target image of the current frame, and the corresponding image features in the target feature library can be replaced with target image features according to a preset replacement rule; if the target occlusion degree score is greater than the preset occlusion threshold, it indicates that the current target is severely occluded in the target image of the current frame, and there is no need to update the target image features to the target feature library.

[0062] In an optional embodiment of the present invention, updating the target feature library according to the target image features using a preset replacement rule may include: if it is determined that there are both positive and negative sample features in the target feature library, replacing the negative sample features in the target feature library with the target image features; if it is determined that there are only positive sample features in the target feature library, updating the target feature library according to the correlation between the positive sample features if the target image features are positive sample features; and replacing the positive sample features last added to the target feature library with the target image features if the target image features are negative sample features.

[0063] Positive sample features can be sample features in the target sample library whose target reliability score is greater than or equal to a preset reliability threshold. Negative sample features can be sample features in the target sample library whose target reliability score is less than the preset reliability threshold. The degree of correlation between positive sample features can be the degree of correlation between positive sample feature attributes in the target feature library.

[0064] It is understood that when the target reliability score is greater than or equal to the preset reliability threshold, the target image feature is determined to be a positive sample feature; when the target reliability score is less than the preset reliability threshold, the target image feature is determined to be a negative sample feature.

[0065] Therefore, when updating the target feature library based on target image features using preset replacement rules, the image features included in the target feature library can be assessed first. If both positive and negative sample features exist in the target feature library, the negative sample features can be replaced using the target image features. If only positive sample features exist in the target feature library, the library can be updated based on the type of the target image features. If the target image feature is determined to be a positive sample feature, the correlation between positive sample features can be calculated, and the target feature library can be updated accordingly, ensuring that similar target sample features are prioritized for replacement. If the target image feature is determined to be a negative sample feature, it can replace the last positive sample feature added to the target feature library. It should be noted that only one negative sample feature can exist in the target feature library.

[0066] like Figure 3 As shown, assume that the target feature library contains {f1, f...} i1 f i2 f i3 f i4 Five image features are listed in sequence. Here, f1 indicates that the image feature was obtained by feature extraction from the first frame of the video sequence. i1 This indicates that the image features were obtained by feature extraction from the i1th frame of the video sequence. i2 This indicates that the image feature was obtained by feature extraction from the i2th frame of the video sequence. i3 This indicates that the image features were obtained by feature extraction from the i3rd frame of the video sequence. i4 This indicates that the image features were obtained by feature extraction from the i4th frame of the video sequence.

[0067] It should be noted that this embodiment of the invention does not limit the number of image features in the target feature library. However, it is understood that as the number of image features in the target feature library increases, the computational burden on the network also increases accordingly. Therefore, to ensure the efficiency of image segmentation, the number of image features in the target feature library can be limited. This embodiment of the invention limits the number of image features in the target feature library to remain constant; therefore, when adding target image features to the target feature library, existing image features in the target feature library can be replaced. If the number of image features in the target feature library is not limited, target image features can be directly added to the target feature library.

[0068] In a specific example, if negative sample features exist in the target feature library, then regardless of whether the target reliability score is less than the preset reliability threshold, the negative sample features in the target feature library will be replaced first; if the target feature library contains only positive sample features, then the positive sample features that need to be replaced in the target feature library can be determined according to the type of target image features, thereby realizing dynamic updating of the target feature library.

[0069] In a specific example, the distance between positive sample image features in the target feature library can be calculated to obtain a distance matrix. Further, the image feature pairs with the smallest distance can be selected from the distance matrix, and then the distances corresponding to these smallest image feature pairs can be evaluated.

[0070] In an optional embodiment of the present invention, updating the target feature library based on the correlation between the positive sample features may include: calculating the distance between each of the positive sample features to obtain a distance matrix of each of the target sample features in the target feature library; filtering the distance matrix to obtain the target sample feature distance and the first positive sample feature pair corresponding to the target sample feature distance; and updating the target feature library based on the target sample feature distance and the first positive sample feature pair corresponding to the target sample feature distance.

[0071] The distance matrix can be a matrix composed of the distances between each positive sample feature. The target sample feature distance can be a distance data point in the distance matrix. The first positive sample feature pair can be the positive sample image feature pair corresponding to the target sample feature distance.

[0072] In this embodiment of the invention, when updating the target feature library based on the correlation between positive sample features, the distance between each positive sample feature can first be calculated, and a distance matrix of each target sample feature in the target feature library can be generated based on the distance between each positive sample feature. Further, the distance matrix can be filtered to obtain the target sample feature distance and the first positive sample feature pair corresponding to the target sample feature distance. For example, the minimum distance can be selected from the distance matrix as the target sample feature distance. After obtaining the target sample feature distance and the first positive sample feature pair corresponding to the target sample feature distance, the target feature library can be updated based on the target sample feature distance and the first positive sample feature pair corresponding to the target sample feature distance.

[0073] In an optional embodiment of the present invention, updating the target feature library based on the target sample feature distance and the first positive sample feature pair corresponding to the target sample feature distance may include: when it is determined that the target sample feature distance is less than a preset sample feature distance, replacing the target replacement sample feature in the first positive sample feature pair with the target image feature; wherein the feature update time of the target replacement sample feature in the first positive sample feature pair is earlier than the feature update time of the non-target replacement sample feature in the first positive sample feature pair; when it is determined that the target sample feature distance is greater than or equal to the preset sample feature distance, calculating the frame interval between each of the positive sample features; filtering the frame intervals between adjacent positive sample features to obtain the target frame interval and the second positive sample feature pair corresponding to the target frame interval; replacing the target replacement sample feature in the second positive sample feature pair in the target feature library with the target image feature; wherein the timestamp of the target replacement sample feature in the second positive sample feature pair is larger than the timestamp of the non-target replacement sample feature in the second positive sample feature pair.

[0074] The preset sample feature distance can be a threshold value for the distance between target sample features in a pre-set target feature library. The frame interval between positive sample features can be the number of frames between each positive sample feature in the video sequence. The target replacement sample feature in the first positive sample feature pair can be the sample feature to be replaced by the target image feature in the first positive sample feature pair. The non-target replacement sample feature in the first positive sample feature pair can be any sample feature in the first positive sample feature pair other than the target replacement sample feature. The feature update time can be the time when the target sample feature is stored in the target feature library. The target frame interval can be the frame interval between two positive sample features. The second positive sample feature pair corresponding to the target frame interval can be two positive sample features corresponding to the target frame interval. The timestamp can be the timestamp of the video frame corresponding to the target sample feature in the target feature library. The target replacement sample feature in the second positive sample feature pair can be the sample feature to be replaced by the target image feature in the second positive sample feature pair. The non-target replacement sample feature in the second positive sample feature pair can be any sample feature in the second positive sample feature pair other than the target replacement sample feature.

[0075] In this embodiment of the invention, when updating the target feature library based on the target sample feature distance and the first positive sample feature pair corresponding to the target sample feature distance, the target sample feature distance can first be determined. If the target sample feature distance is less than a preset sample feature distance, the target image feature can be used to replace the target sample feature with the earlier update time in the first positive sample feature pair; if the target sample feature distance is greater than or equal to the preset sample feature distance, the frame interval between positive sample features in the target feature library can be further calculated. For example, the frame interval between positive sample features in the target feature library can be {i1-1,i2-i1,i3-i2,i4-i3}. After obtaining the frame interval between positive sample features in the target feature library, the smallest frame interval can be used as the target frame interval, and the two target sample features with the smallest frame interval can be used as the second positive sample feature pair corresponding to the target frame interval. Furthermore, the target image feature can be used to replace the target sample feature corresponding to the later video frame in the second positive sample feature pair, thereby increasing the diversity of target sample features in the target feature library and improving the accuracy of image segmentation processing. For example, assuming i4-i3 is the target frame interval, the target image features can be used to replace the target sample features f corresponding to i4. i4

[0076] This invention employs a target feature library to perform image segmentation on the target image of the current frame, obtaining the target image features of the current target, and determining the target reliability score and target occlusion score of the target image features. Further, it determines whether the target reliability score is less than a preset reliability threshold. If the target reliability score is less than the preset reliability threshold, the target feature library is updated according to the target image features using a preset replacement rule; if the target reliability score is greater than or equal to the preset reliability threshold, the target feature library is updated according to the target occlusion score and the target image features. This scheme, by dynamically updating the target feature library, ensures that the feature library always stores high-quality image features highly adapted to the current video segmentation scenario, solving the problem of poor image feature quality in the feature library in existing technologies, thereby improving the accuracy of video target segmentation.

[0077] The collection, storage, use, processing, transmission, provision, and disclosure of user personal information (such as facial information, image information, etc.) involved in the technical solution disclosed herein comply with the provisions of relevant laws and regulations and do not violate public order and good morals.

[0078] It should be noted that all information (including but not limited to user device information, user personal information, etc.) and data (including but not limited to data used for display, data used for analysis, etc.) involved in this disclosure are information and data authorized by the user or fully authorized by all parties, and the collection, use and processing of the relevant data comply with the relevant laws, regulations and standards of the relevant regions.

[0079] It should be noted that any arrangement or combination of the technical features in the above embodiments also falls within the protection scope of this invention.

[0080] Example 3

[0081] Figure 4 This is a schematic diagram of a feature library updating device provided in Embodiment 3 of the present invention, as shown below. Figure 4 As shown, the device includes: a target image feature acquisition module 310, a target image feature scoring and determination module 320, and a target feature library update module 330, wherein:

[0082] The target image feature acquisition module 310 is used to perform image segmentation processing on the target image of the current frame to obtain the target image features of the current target.

[0083] The target image feature scoring determination module 320 is used to determine the target reliability score and target occlusion degree score of the target image features.

[0084] The target feature library update module 330 is used to update the target feature library based on the target reliability score and the target occlusion score.

[0085] The target feature library is used to perform image segmentation processing on the target image of the current frame.

[0086] This invention employs a target feature library to perform image segmentation on the target image of the current frame, thereby obtaining the target image features of the current target. After obtaining the target image features of the current target, a target reliability score and a target occlusion score are determined for the target image features, and the target feature library is updated based on these scores. This scheme, by dynamically updating the target feature library, ensures that the feature library always stores high-quality image features that are highly adapted to the current video segmentation scenario. This solves the problem of poor image feature quality in the feature library in existing technologies, improves the quality of image features in the feature library, and thus improves the accuracy of video target segmentation.

[0087] Optionally, the target feature library update module 330 is specifically used to: update the target feature library according to the target image features by means of a preset replacement rule when the target reliability score is determined to be less than a preset reliability threshold; and update the target feature library according to the target occlusion degree score and the target image features when the target reliability score is determined to be greater than or equal to the preset reliability threshold.

[0088] Optionally, the target feature library update module 330 is further configured to: update the target feature library according to the target image features using the preset replacement rule when the target occlusion score is determined to be less than or equal to a preset occlusion threshold; and prohibit updating the target feature library when the target occlusion score is determined to be greater than the preset occlusion threshold.

[0089] Optionally, the target feature library update module 330 is further configured to: replace the negative sample features in the target feature library with the target image features when it is determined that both positive and negative sample features exist in the target feature library; update the target feature library according to the correlation between the positive sample features if the target image features are positive sample features, and replace the last positive sample features added to the target feature library with the target image features if the target image features are negative sample features.

[0090] Optionally, the target feature library update module 330 is further configured to: calculate the distance between each of the positive sample features to obtain a distance matrix of each of the target sample features in the target feature library; filter the distance matrix to obtain the target sample feature distance and the first positive sample feature pair corresponding to the target sample feature distance; and update the target feature library according to the target sample feature distance and the first positive sample feature pair corresponding to the target sample feature distance.

[0091] Optionally, the target feature library update module 330 is further configured to: when it is determined that the target sample feature distance is less than a preset sample feature distance, replace the target replacement sample feature in the first positive sample feature pair with the target image feature; wherein the feature update time of the target replacement sample feature in the first positive sample feature pair is earlier than the feature update time of the non-target replacement sample feature in the first positive sample feature pair; when it is determined that the target sample feature distance is greater than or equal to the preset sample feature distance, calculate the frame interval between each of the positive sample features; filter the frame interval between adjacent positive sample features to obtain a target frame interval and a second positive sample feature pair corresponding to the target frame interval; replace the target replacement sample feature in the second positive sample feature pair in the target feature library with the target image feature; wherein the timestamp of the target replacement sample feature in the second positive sample feature pair is larger than the timestamp of the non-target replacement sample feature in the second positive sample feature pair.

[0092] The aforementioned feature library update apparatus can execute the feature library update method provided in any embodiment of the present invention, and has the corresponding functional modules and beneficial effects of the method. Technical details not described in detail in this embodiment can be found in the feature library update method provided in any embodiment of the present invention.

[0093] Since the feature library updating apparatus described above is an apparatus capable of executing the feature library updating method in the embodiments of the present invention, those skilled in the art can understand the specific implementation and various variations of the feature library updating apparatus in this embodiment based on the feature library updating method described in the embodiments of the present invention. Therefore, how the feature library updating apparatus implements the feature library updating method in the embodiments of the present invention will not be described in detail here. Any apparatus used by those skilled in the art to implement the feature library updating method in the embodiments of the present invention falls within the scope of protection of this application.

[0094] Example 4

[0095] Figure 5A schematic diagram of an electronic device 10 that can be used to implement embodiments of the present invention is shown. The electronic device is intended to represent various forms of digital computers, such as laptop computers, desktop computers, workstations, personal digital assistants, servers, blade servers, mainframe computers, and other suitable computers. The electronic device can also represent various forms of mobile devices, such as personal digital processors, cellular phones, smartphones, wearable devices (e.g., helmets, glasses, watches, etc.), and other similar computing devices. The components shown herein, their connections and relationships, and their functions are merely illustrative and are not intended to limit the implementation of the invention described and / or claimed herein.

[0096] like Figure 5 As shown, the electronic device 10 includes at least one processor 11 and a memory, such as a read-only memory (ROM) 12 or a random access memory (RAM) 13, communicatively connected to the at least one processor 11. The memory stores computer programs executable by the at least one processor. The processor 11 can perform various appropriate actions and processes based on the computer program stored in the ROM 12 or loaded from storage unit 18 into the RAM 13. The RAM 13 may also store various programs and data required for the operation of the electronic device 10. The processor 11, ROM 12, and RAM 13 are interconnected via a bus 14. An input / output (I / O) interface 15 is also connected to the bus 14.

[0097] Multiple components in electronic device 10 are connected to I / O interface 15, including: input unit 16, such as keyboard, mouse, etc.; output unit 17, such as various types of displays, speakers, etc.; storage unit 18, such as disk, optical disk, etc.; and communication unit 19, such as network card, modem, wireless transceiver, etc. Communication unit 19 allows electronic device 10 to exchange information / data with other devices through computer networks such as the Internet and / or various telecommunications networks.

[0098] Processor 11 can be a variety of general-purpose and / or special-purpose processing components with processing and computing capabilities. Some examples of processor 11 include, but are not limited to, a central processing unit (CPU), a graphics processing unit (GPU), various special-purpose artificial intelligence (AI) computing chips, various processors running machine learning model algorithms, a digital signal processor (DSP), and any suitable processor, controller, microcontroller, etc. Processor 11 performs the various methods and processes described above, such as the feature library update method.

[0099] In some embodiments, the feature library update method may be implemented as a computer program tangibly contained in a computer-readable storage medium, such as storage unit 18. In some embodiments, part or all of the computer program may be loaded and / or installed on electronic device 10 via ROM 12 and / or communication unit 19. When the computer program is loaded into RAM 13 and executed by processor 11, one or more steps of the feature library update method described above may be performed. Alternatively, in other embodiments, processor 11 may be configured to perform the feature library update method by any other suitable means (e.g., by means of firmware).

[0100] Optionally, the feature library update method may include: performing image segmentation processing on the target image of the current frame to obtain target image features of the current target; determining the target reliability score and the target occlusion score of the target image features; and updating the target feature library according to the target reliability score and the target occlusion score; wherein the target feature library is used to perform image segmentation processing on the target image of the current frame.

[0101] Various embodiments of the systems and techniques described above herein can be implemented in digital electronic circuit systems, integrated circuit systems, field-programmable gate arrays (FPGAs), application-specific integrated circuits (ASICs), application-specific standard products (ASSPs), systems-on-a-chip (SoCs), payload-programmable logic devices (CPLDs), computer hardware, firmware, software, and / or combinations thereof. These various embodiments may include implementations in one or more computer programs that can be executed and / or interpreted on a programmable system including at least one programmable processor, which may be a dedicated or general-purpose programmable processor, capable of receiving data and instructions from a storage system, at least one input device, and at least one output device, and transmitting data and instructions to the storage system, the at least one input device, and the at least one output device.

[0102] Computer programs used to implement the methods of the present invention may be written in any combination of one or more programming languages. These computer programs may be provided to a processor of a general-purpose computer, a special-purpose computer, or other programmable data processing device, such that when executed by the processor, the computer programs cause the functions / operations specified in the flowcharts and / or block diagrams to be performed. The computer programs may be executed entirely on a machine, partially on a machine, or as a standalone software package, partially on a machine and partially on a remote machine, or entirely on a remote machine or server.

[0103] In the context of this invention, a computer-readable storage medium can be a tangible medium that may contain or store a computer program for use by or in conjunction with an instruction execution system, apparatus, or device. A computer-readable storage medium may include, but is not limited to, electronic, magnetic, optical, electromagnetic, infrared, or semiconductor systems, apparatus, or devices, or any suitable combination thereof. Alternatively, a computer-readable storage medium may be a machine-readable signal medium. More specific examples of machine-readable storage media include electrical connections based on one or more wires, portable computer disks, hard disks, random access memory (RAM), read-only memory (ROM), erasable programmable read-only memory (EPROM or flash memory), optical fibers, portable compact disk read-only memory (CD-ROM), optical storage devices, magnetic storage devices, or any suitable combination thereof.

[0104] To provide interaction with a user, the systems and techniques described herein can be implemented on an electronic device having: a display device (e.g., a CRT (cathode ray tube) or LCD (liquid crystal display) monitor) for displaying information to the user; and a keyboard and pointing device (e.g., a mouse or trackball) through which the user provides input to the electronic device. Other types of devices can also be used to provide interaction with the user; for example, feedback provided to the user can be any form of sensory feedback (e.g., visual feedback, auditory feedback, or tactile feedback); and input from the user can be received in any form (including sound input, voice input, or tactile input).

[0105] The systems and technologies described herein can be implemented in computing systems that include backend components (e.g., as data servers), or computing systems that include middleware components (e.g., application servers), or computing systems that include frontend components (e.g., user computers with graphical user interfaces or web browsers through which users can interact with implementations of the systems and technologies described herein), or any combination of such backend, middleware, or frontend components. The components of the system can be interconnected via digital data communication of any form or medium (e.g., communication networks). Examples of communication networks include local area networks (LANs), wide area networks (WANs), blockchain networks, and the Internet.

[0106] A computing system can include clients and servers. Clients and servers are generally located far apart and typically interact through communication networks. The client-server relationship is created by computer programs running on the respective computers and having a client-server relationship with each other. The server can be a cloud server, also known as a cloud computing server or cloud host, which is a hosting product within the cloud computing service system to address the shortcomings of traditional physical hosts and VPS services, such as high management difficulty and weak business scalability.

[0107] It should be understood that the various forms of processes shown above can be used to rearrange, add, or delete steps. For example, the steps described in this disclosure can be executed in parallel, sequentially, or in different orders, as long as the desired result of the technical solution disclosed in this disclosure can be achieved, and this is not limited herein.

[0108] The specific embodiments described above do not constitute a limitation on the scope of protection of this disclosure. Those skilled in the art should understand that various modifications, combinations, sub-combinations, and substitutions can be made according to design requirements and other factors. Any modifications, equivalent substitutions, and improvements made within the spirit and principles of this disclosure should be included within the scope of protection of this disclosure.

Claims

1. A feature library update method, characterized in that, include: Perform image segmentation processing on the target image of the current frame to obtain the target image features of the current target; Determine the target reliability score and target occlusion score of the target image features; The target feature library is updated based on the target reliability score and the target occlusion score. The target feature library is used to perform image segmentation processing on the target image of the current frame.

2. The method according to claim 1, characterized in that, The step of updating the target feature library based on the target reliability score and the target occlusion score includes: If the target reliability score is determined to be less than a preset reliability threshold, the target feature library is updated according to the target image features using a preset replacement rule; If the target reliability score is determined to be greater than or equal to a preset reliability threshold, the target feature library is updated based on the target occlusion score and the target image features.

3. The method according to claim 2, characterized in that, The step of updating the target feature library based on the target occlusion score and the target image features includes: If the target occlusion score is determined to be less than or equal to a preset occlusion threshold, the target feature library is updated according to the target image features based on the preset replacement rule. If the target occlusion score is determined to be greater than a preset occlusion threshold, updating the target feature library is prohibited.

4. The method according to claim 2 or 3, characterized in that, The step of updating the target feature library according to the target image features using a preset replacement rule includes: If it is determined that there are both positive and negative sample features in the target feature library, the negative sample features in the target feature library are replaced by the target image features. If it is determined that only the positive sample features exist in the target feature library, and if the target image feature is a positive sample feature, then the target feature library is updated according to the degree of correlation between the positive sample features; if the target image feature is a negative sample feature, then the target image feature replaces the positive sample feature last added to the target feature library.

5. The method according to claim 4, characterized in that, The step of updating the target feature library based on the correlation between the positive sample features includes: Calculate the distance between each of the positive sample features to obtain the distance matrix of each of the target sample features in the target feature library; The distance matrix is ​​filtered to obtain the target sample feature distance and the first positive sample feature pair corresponding to the target sample feature distance; The target feature library is updated based on the target sample feature distance and the first positive sample feature pair corresponding to the target sample feature distance.

6. The method according to claim 5, characterized in that, The step of updating the target feature library based on the target sample feature distance and the first positive sample feature corresponding to the target sample feature distance includes: If the target sample feature distance is determined to be less than the preset sample feature distance, the target replacement sample feature in the first positive sample feature pair is replaced according to the target image feature; wherein, the feature update time of the target replacement sample feature in the first positive sample feature pair is earlier than the feature update time of the non-target replacement sample feature in the first positive sample feature pair. If the target sample feature distance is determined to be greater than or equal to the preset sample feature distance, the frame interval between each of the positive sample features is calculated; The frame intervals between adjacent positive sample features are filtered to obtain a target frame interval and a second positive sample feature pair corresponding to the target frame interval; Replace the target replacement sample feature in the second positive sample feature pair in the target feature library with the target image feature; wherein the timestamp of the target replacement sample feature in the second positive sample feature pair is larger than the timestamp of the non-target replacement sample feature in the second positive sample feature pair.

7. A feature library updating device, characterized in that, include: The target image feature acquisition module is used to perform image segmentation processing on the target image of the current frame to obtain the target image features of the current target. The target image feature scoring and determination module is used to determine the target reliability score and target occlusion score of the target image features. The target feature library update module is used to update the target feature library based on the target reliability score and the target occlusion score; The target feature library is used to perform image segmentation processing on the target image of the current frame.

8. An electronic device, characterized in that, The electronic device includes: At least one processor; and A memory communicatively connected to the at least one processor; wherein, The memory stores a computer program that is executed by the at least one processor to enable the at least one processor to perform the feature library update method according to any one of claims 1-6.

9. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores computer instructions that cause a processor to execute the feature library update method according to any one of claims 1-6.

10. A computer program product comprising a computer program / instructions, wherein, When the computer program / instructions are executed by the processor, they implement the feature library update method according to any one of claims 1-6.

Images