Target scene algorithm issuing method and device based on visual large model and computer equipment

Abstract

The invention relates to a target scene algorithm issuing method and device based on a visual large model and computer equipment, and the method comprises the steps: obtaining a current label set and a last analysis label set based on equipment with a built-in visual large model; the equipment is associated with various scene algorithms; determining a tag data set of the equipment according to sub-tags in the tag set of this time and sub-tags in the tag set of last analysis; performing matching according to the label data set and a preset first corresponding relation list to obtain a matching result; issuing of a target scene algorithm is completed according to a matching result; the first corresponding relation list has a mapping relation between the scene algorithm and the label. Through application of the method and the device, the problem of limitation of an algorithm use scene in a terminal in related technologies is solved, different scene algorithms are issued for different scene algorithms, interference of scene algorithm environment factors can be effectively avoided, the accuracy of scene algorithm issuing under scene conversion is improved, and the recall rate is reduced.

Description

Technical Field

[0001] The present application relates to the field of computer vision technology, and in particular to a method, apparatus, and computer device for issuing a target scene algorithm based on a large visual model. Background Art

[0002] With the development of artificial intelligence (AI), algorithms have been widely applied across industries. For example, camera algorithms, based on video frame extraction and frame analysis, have enabled the development of algorithms such as fire status notification, mask notification, intrusion detection, and vehicle notification. These algorithms are widely used in business monitoring and management across various sectors, including commercial, industrial, and transportation. However, in practice, algorithms are often limited in their use cases. For example, currently, most algorithms in the industry are only applicable to a single scenario. For example, algorithms trained on outdoor scenes will significantly degrade in various algorithm metrics when applied indoors. Algorithms trained on indoor scenes also perform poorly in outdoor settings. Furthermore, algorithms trained on indoor and outdoor data, due to their ability to accommodate diverse scenarios, are less capable than algorithms trained on a single scenario. For example, an electric vehicle notification algorithm trained for elevator scenarios suffers from reduced accuracy and recall when applied to corridors and doorways due to interference from environmental factors such as lighting.

[0003] There is currently no effective solution to the problem that the algorithm usage scenarios in the terminal are limited and the accuracy of the algorithm is low due to environmental factors in related technologies. Summary of the Invention

[0004] In this embodiment, a method, apparatus, and computer device for sending a target scene algorithm based on a large visual model are provided to solve the problem in related technologies that the algorithm usage scenario in the terminal is limited and the accuracy of the algorithm sending is low due to environmental factors.

[0005] First, in this embodiment, a method for issuing a target scene algorithm based on a large visual model is provided, including:

[0006] Based on a device with a built-in large visual model, the current tag set and the last analyzed tag set are obtained; the device is associated with various scene algorithms;

[0007] Determining a tag data set for the device based on the subtags in the current tag set and the subtags in the last analyzed tag set;

[0008] Matching is performed based on the tag data set and a preset first correspondence list to obtain a matching result; and the target scene algorithm is issued based on the matching result; the first correspondence list contains a mapping relationship between the scene algorithm and the sub-tag.

[0009] In some embodiments, obtaining the tag set based on a device with a built-in visual large model includes:

[0010] Capturing a video stream based on a device with a built-in visual model, and extracting picture frames from the video stream at preset time intervals;

[0011] The picture frame is subjected to picture frame label analysis of a large visual model, and the current label set of the picture frame is output in a set form.

[0012] In some embodiments, determining the tag data set of the device based on the current tag set and the last analyzed tag set includes:

[0013] Compare the subtags in the current tag set with the subtags in the last analyzed tag set to determine whether there are different subtags;

[0014] If there are different sub-tags, the sub-tags in the last analyzed tag set are reset to the sub-tags in the current tag set;

[0015] If there is no difference sub-tag, the tag data set of the device is updated based on the current tag set.

[0016] In some embodiments, matching the tag data set with a preset first correspondence list to obtain a matching result includes:

[0017] Determining a scene algorithm corresponding to each sub-tag according to the tag data set and a preset first correspondence list;

[0018] If multiple sub-tags correspond to one scene algorithm, the corresponding scene algorithm is determined as the target scene algorithm to be used by the device;

[0019] If multiple subtags correspond to multiple scene algorithms, the score of each corresponding scene algorithm is calculated according to a preset scoring formula, and the scene algorithm with the highest score is determined as the target scene algorithm to be used by the device;

[0020] If multiple sub-tags do not correspond to scene algorithms, the universal scene algorithm is determined as the target scene algorithm that the device should use.

[0021] In some embodiments, the score formula is weight coefficient k1×label 1+weight coefficient k2×label 2+…weight coefficient kn×label n.

[0022] In some embodiments, the method further comprises:

[0023] According to the label data set and a preset second correspondence list, determine the business product corresponding to the label data set in the second correspondence list, and mark the determined business product as the target business product of the device; the second correspondence list has a mapping relationship between the business product and the sub-label.

[0024] In some embodiments, the method further comprises:

[0025] Promotion is performed for users corresponding to the device based on the target business product.

[0026] In a second aspect, in this embodiment, a device for issuing a target scene algorithm based on a large visual model is provided, comprising: an acquisition module, a processing module, and an issuing module;

[0027] The acquisition module is used to acquire the current tag set and the last analyzed tag set based on a device with a built-in visual large model; the device is associated with various scene algorithms;

[0028] The processing module is configured to determine the tag data set of the device based on the subtags in the current tag set and the subtags in the last analyzed tag set;

[0029] The sending module is used to match the tag data set, the scene algorithm and the preset first correspondence list to obtain a matching result; and complete the sending of the target scene algorithm according to the matching result; the first correspondence list has a mapping relationship between the scene algorithm and the sub-tag.

[0030] In a third aspect, a computer device is provided in this embodiment, comprising a memory, a processor, and a computer program stored on the memory and executable on the processor. When the processor executes the computer program, the method for delivering the target scene algorithm based on the visual large model described in the first aspect above is implemented.

[0031] In a fourth aspect, a storage medium is provided in this embodiment, on which a computer program is stored. When the program is executed by a processor, the method for issuing a target scene algorithm based on a large visual model described in the first aspect above is implemented.

[0032] Compared with the related art, the method, device and computer equipment for issuing the target scene algorithm based on the visual big model provided in this embodiment obtain the current label set and the last analysis label set through a device with a built-in visual big model; the device is associated with various scene algorithms; the label data set of the device is determined according to the sub-labels in the current label set and the sub-labels in the last analysis label set; matching is performed according to the label data set and the preset first correspondence list to obtain the matching result; and the target scene algorithm is issued according to the matching result; the first correspondence list has a mapping relationship between the scene algorithm and the label, which solves the problem of the limitation of the algorithm usage scenario in the terminal in the related art, and dynamically matches the corresponding scene algorithm according to the label, thereby realizing the issuance of different scene algorithms for different scene algorithms, which can effectively avoid the scene algorithm from being interfered with by environmental factors such as weather and light, improve the accuracy of the scene algorithm issuance under scene conversion, and reduce the recall rate.

[0033] The details of one or more embodiments of the present application are set forth in the following drawings and description to make other features, objects, and advantages of the present application more readily apparent. BRIEF DESCRIPTION OF THE DRAWINGS

[0034] The drawings described herein are used to provide a further understanding of the present application and constitute a part of the present application. The illustrative embodiments of the present application and their descriptions are used to explain the present application and do not constitute an improper limitation on the present application. In the drawings:

[0035] Figure 1 This is a hardware structure block diagram of a terminal device for a method for issuing a target scene algorithm based on a large visual model provided in one embodiment of the present application;

[0036] Figure 2 This is a flowchart of a method for issuing a target scene algorithm based on a large visual model provided by an embodiment of the present application;

[0037] Figure 3 is a flowchart of step S220;

[0038] Figure 4 is a flowchart of step S230;

[0039] Figure 5 This is a flowchart of a method for issuing a target scene algorithm based on a large visual model provided by another embodiment of the present application;

[0040] Figure 6 This is a structural block diagram of a device for issuing a target scene algorithm based on a large visual model provided in one embodiment of the present application.

[0041] In the figure: 102, processor; 104, memory; 106, transmission device; 108, input and output device; 210, acquisition module; 220, processing module; 230, sending module. DETAILED DESCRIPTION

[0042] In order to more clearly understand the purpose, technical solutions and advantages of the present application, the present application is described and illustrated below in conjunction with the accompanying drawings and embodiments.

[0043] Unless otherwise defined, technical or scientific terms used in this application shall have the ordinary meanings as understood by persons of ordinary skill in the art to which this application belongs. The terms "a," "an," "the," "these," and similar expressions in this application do not denote limitations on quantity and may be singular or plural. The terms "comprise," "include," "have," and any variations thereof, as used in this application, are intended to cover non-exclusive inclusions. For example, a process, method, system, product, or device comprising a series of steps or modules (units) is not limited to the listed steps or modules (units) but may include unlisted steps or modules (units) or other steps or modules (units) inherent to the process, method, product, or device. The terms "connected," "connected," "coupled," and similar expressions used in this application are not limited to physical or mechanical connections but may include electrical connections, whether direct or indirect. As used in this application, "plurality" means two or more. "And / or" describes an association between associated objects, indicating that three possible relationships exist. For example, "A and / or B" can mean: A exists alone; A and B exist simultaneously; or B exists alone. Generally, the character " / " indicates that the objects in the preceding and following relationship are in an "or" relationship. The terms "first", "second", "third", etc. involved in this application are only used to distinguish similar objects and do not represent a specific ordering of the objects.

[0044] The method embodiment provided in this embodiment can be executed in a terminal, a computer or a similar computing device. For example, running on a terminal, Figure 1 This is a hardware block diagram of the terminal of the method for issuing the target scene algorithm based on the visual large model of this embodiment. Figure 1 As shown, the terminal may include one or more ( Figure 1 The processor 102 (only one is shown) and a memory 104 for storing data, wherein the processor 102 may include but is not limited to a processing device such as a microprocessor MCU or a programmable logic device FPGA. The terminal may also include a transmission device 106 for communication functions and an input / output device 108. It will be understood by those skilled in the art that Figure 1The structure shown is only for illustration and does not limit the structure of the above terminal. Figure 1 More or fewer components than shown, or with Figure 1 Different configurations shown.

[0045] The memory 104 can be used to store computer programs, for example, software programs and modules of application software, such as the computer program corresponding to the method for issuing the target scene algorithm based on the large visual model in this embodiment. The processor 102 executes various functional applications and data processing by running the computer program stored in the memory 104, that is, implementing the above-mentioned method. The memory 104 may include a high-speed random access memory and may also include a non-volatile memory, such as one or more magnetic storage devices, flash memory, or other non-volatile solid-state memory. In some instances, the memory 104 may further include a memory remotely located relative to the processor 102, and these remote memories may be connected to the terminal via a network. Examples of the above-mentioned network include, but are not limited to, the Internet, an intranet, a local area network, a mobile communication network, and combinations thereof.

[0046] Transmission device 106 is used to receive or transmit data via a network. This network may include a wireless network provided by the terminal's communications provider. In one embodiment, transmission device 106 includes a network interface controller (NIC), which can be connected to other network devices via a base station to enable communication with the Internet. In another embodiment, transmission device 106 may be a radio frequency (RF) module, which is used to communicate with the Internet wirelessly.

[0047] In this embodiment, a method for issuing a target scene algorithm based on a large visual model is provided. Figure 2 This is a flow chart of the method for issuing a target scene algorithm based on a large visual model in this embodiment. Figure 2 As shown, the process includes the following steps:

[0048] Step S210: Based on a device with a built-in visual large model, the current tag set and the last analyzed tag set are obtained; the device is associated with various scene algorithms;

[0049] Step S220, determining the device's tag data set based on the sub-tags in the current tag set and the sub-tags in the last analyzed tag set;

[0050] Step S230, matching is performed based on the tag data set and the preset first correspondence list to obtain a matching result; and the target scene algorithm is issued based on the matching result; the first correspondence list has a mapping relationship between the scene algorithm and the tag.

[0051] Specifically, devices refer to cameras and camcorders equipped with built-in vision foundation models and various scene algorithms. These can also be considered terminals or edge devices. Vision foundation models (FFMs) are deep learning models trained on large-scale data with powerful general visual understanding capabilities. They typically have a large number of parameters (over a billion) and can handle a variety of visual tasks. These models include, but are not limited to, multimodal vision foundation models, pure vision foundation models, generative vision models, and industry-specific vision foundation models. Scene algorithms include, but are not limited to, fire recognition algorithms (outdoor), fire recognition algorithms (indoor), fire recognition algorithms (forest), fire recognition algorithms (general), handwriting recognition, and face recognition. Obtaining the current label set and the last analyzed label set is a prerequisite. The last analyzed label set refers to the label data set obtained after processing the previous current label set. It can also be the label set obtained by extracting the previous frame, without limitation. The current label set refers to the set of sub-labels obtained by processing the current frame captured by the device using the FFM. A frame is the smallest slice of a video image, i.e., a frozen image of the video image.

[0052] Among them, the device's label data set is obtained by comparing the sub-labels in the current label set with the sub-labels in the last analyzed label set. The comparison between the sub-labels in the last analyzed label set and the sub-labels in the current label set is used to complete the credibility verification of whether the device's screen has changed, thereby increasing the reliability of the device's label data set, avoiding inaccurate labels caused by interference from environmental factors such as weather, light, and moving objects, and reducing the occurrence of false alarms and AI hallucinations.

[0053] The first correspondence list is pre-set and contains mappings between scene algorithms and labels. The label data set is then matched against the first correspondence list to obtain a matching result. The matching result is the target scene algorithm corresponding to the device, which is then distributed to the device, improving the accuracy of scene algorithm distribution during scene transitions and reducing recall.

[0054] In related technologies, the algorithms in devices are limited in their use cases. For example, currently, most algorithms in the industry are only applicable to one scenario. For example, algorithms trained for outdoor scenarios will significantly reduce various algorithm indicators when used indoors. Algorithms trained for indoor scenarios also perform poorly in outdoor scenarios. Furthermore, algorithms trained using indoor and outdoor training data are less capable than algorithms trained for a single scenario, as they can accommodate different scenarios. For example, an electric vehicle delivery algorithm trained for elevator scenarios suffers from reduced accuracy and recall when used in corridors and doorways due to interference from environmental factors such as lighting. In this embodiment, based on a device with a built-in visual large model, the current label set and the last analyzed label set are obtained; the device is associated with various scene algorithms; the label data set of the device is determined based on the sub-labels in the current label set and the sub-labels in the last analyzed label set; matching is performed based on the label data set and the preset first correspondence list to obtain a matching result; and the target scene algorithm is sent based on the matching result; the first correspondence list has a mapping relationship between scene algorithms and labels, which solves the problem of algorithm usage scenario limitations in the terminal in related technologies, and dynamically matches the corresponding scene algorithms according to the labels, thereby realizing the sending of different scene algorithms for different scene algorithms, which can effectively avoid the scene algorithm from being interfered with by environmental factors such as weather and light, improve the accuracy of the scene algorithm sent under scene conversion, and reduce the recall rate.

[0055] The following is a detailed description of each of the above steps:

[0056] In some embodiments, obtaining the tag set based on a device with a built-in visual large model includes the following steps:

[0057] The device with built-in visual model collects video streams and extracts frames from the video stream at preset time intervals.

[0058] The picture frame is subjected to the picture frame label analysis of the visual large model, and the current label set of the picture frame is output in the form of a set.

[0059] Specifically, before extracting the picture frame, the device can be initialized and a back-end database can be established; the back-end database is used to store various types of data of the above-mentioned method embodiment, including but not limited to video streams, picture frames, this label set and the last analysis label set, label data set, first correspondence list and second correspondence list, etc.

[0060] After initializing the device, it captures the video stream and extracts frames M from it at a preset interval T. The interval T and frame M can be set based on the usage scenario and are not restricted. Frames M are then fed into the visual model for frame label analysis. The resulting label set P is then output as a collection. For example, the current label set P is: Image M {label 1, label 2, label 3}.

[0061] Through this embodiment, the picture frame can be extracted quickly and accurately to complete the acquisition of this tag set.

[0062] In some of these embodiments, Figure 3 As shown, step S220 determines the tag data set of the device based on the current tag set and the last analyzed tag set, including the following steps:

[0063] Step S221: Compare the subtags in the current tag set with the subtags in the last analyzed tag set to determine whether there are different subtags.

[0064] Step S222: If there are different sub-tags, the sub-tags in the last analyzed tag set are reset to the sub-tags in the current tag set;

[0065] Step S223: If there is no different sub-tag, the tag data set of the device is updated based on the current tag set.

[0066] This embodiment can be considered as a step of tag credibility verification and device labeling. Specifically:

[0067] Compare the subtags in the current tag set P with the subtags in the last analyzed tag set P', traverse all subtags in the two sets, and determine whether there are different subtags;

[0068] If the subtags in the current tag set P are not completely consistent with the subtags in the last analyzed tag set P', it is considered that there are discrepant subtags. The subtags in the last analyzed tag set P' are reset to the subtags in the current tag set P, and the device's tag data set is updated based on the reset current tag set. When there are discrepant subtags, it can be assumed that the device's current usage scenario (such as location change, lighting change, or image change) has changed. For example, the device was originally installed indoors, and the current tag set P is {tag 1, tag 2, tag 3}, calling algorithm A for face recognition. Now it is moved outdoors, and the last analyzed tag set P' is {tag 1, tag 2}. There are discrepant subtags. If algorithm A is still used, the recognition effect will be significantly reduced. Therefore, subsequent steps are required to link and issue the corresponding scene algorithm (algorithm B).

[0069] If the subtags in the current tag set P are identical to the subtags in the previously analyzed tag set P', it is assumed that there are no subtag differences. If the device's current usage scenario (e.g., location change, lighting change, or image change) has not changed, the device's tag data set L is updated based on the current tag set P, making it equal to the current tag set P. At this point, the previous scenario algorithm can be directly delivered to the device. For example, if the device was originally installed indoors, the current tag set P was {tag 1, tag 2, tag 3}, and algorithm A was invoked for face recognition. Now, if the device is still indoors and the previous tag set P' was {tag 1, tag 2, tag 3}, there are no subtag differences, so algorithm A is directly delivered, significantly saving computational and scheduling resources.

[0070] Through this embodiment, the process of credibility verification and device labeling is utilized to increase the reliability of the label data set, and the labeling and storage of the sub-labels in the last analyzed label set are avoided, thereby avoiding the situation where the device is disturbed by environmental factors such as weather, light, and moving objects, causing inaccurate labels, greatly reducing the occurrence of false alarms, and solving the AI ​​hallucination problem.

[0071] In some of these embodiments, Figure 4 As shown, step S230 performs matching based on the tag data set and the preset first correspondence list to obtain a matching result, including the following steps:

[0072] Step S231, determining the scene algorithm corresponding to each sub-tag according to the tag data set and the preset first correspondence list;

[0073] Step S232: If multiple sub-tags correspond to one scene algorithm, the corresponding scene algorithm is determined as the target scene algorithm to be used by the device;

[0074] Step S233: If multiple subtags correspond to multiple scene algorithms, the score of each corresponding scene algorithm is calculated according to a preset scoring formula, and the scene algorithm with the highest score is determined as the target scene algorithm to be used by the device;

[0075] Step S234: If the multiple sub-tags do not correspond to a scene algorithm, the universal scene algorithm is determined as the target scene algorithm that the device should use.

[0076] Specifically, the first correspondence list may be as shown in Table 1 (the mapping relationship shown in Table 1 is only for illustration and does not limit the above mapping relationship).

[0077] Table 1

[0078]

[0079] Match the labels in the label data set with the first correspondence list.

[0080] If multiple sub-tags correspond to a scene algorithm, the corresponding scene algorithm is determined as the target scene algorithm that the device should use. For example, if the tags in the tag data set are tag 1 and tag 2, and they correspond to the fire recognition algorithm (outdoor), the fire recognition algorithm (outdoor) is determined as the target scene algorithm that the device should use, and the fire recognition algorithm (outdoor) is issued for fire recognition.

[0081] If multiple sub-tags correspond to multiple scenario algorithms, the scores of each corresponding scenario algorithm are calculated according to a preset scoring formula, and the scenario algorithm with the highest score is determined as the target scenario algorithm to be used by the device. The scoring formula is: weight coefficient k1 × label 1 + weight coefficient k2 × label 2 + ... weight coefficient kn × label n. For example, if the labels in the label data set are label 1, label 2, and label 5, and each label has a score of 1, weight coefficient k1 is 0.1, weight coefficient k2 is 0.1, and weight coefficient k5 is 0.3, then the scenario algorithms corresponding to labels 1 and 2 have a score of 0.2, and the scenario algorithm corresponding to label 5 has a score of 0.3. Therefore, if the scenario algorithm corresponds to the Fire Recognition Algorithm (Forest), the Fire Recognition Algorithm (Forest) is determined as the target scenario algorithm to be used by the device and is issued for fire recognition. The weight coefficient and the score of label 2 can be set based on the usage scenario and are not limited thereto. In other embodiments, the scenario algorithm scores can also be calculated using methods such as pre-trained neural network models.

[0082] If multiple subtags do not correspond to a scenario algorithm, the universal scenario algorithm is determined as the target scenario algorithm for the device. For example, if the tag in the tag data set is tag 7 and corresponds to the Fire Recognition Algorithm (Universal), the Fire Recognition Algorithm (Universal) is determined as the target scenario algorithm for the device and issued for fire recognition.

[0083] This embodiment distributes the corresponding scenario algorithm based on the device's tag data set, ensuring that the scenario algorithm adapts to the usage environment. For multiple tags, different weight coefficients are used to score and sort the tags, selecting the optimal scenario algorithm for dispatch and distribution. This integration of the scenario algorithm with the usage environment prevents environmental factors from interfering with the scenario algorithm, significantly improving the accuracy of scenario algorithm distribution during scenario transitions and reducing recall rates.

[0084] In some of these embodiments, Figure 5 As shown, the method for issuing the target scene algorithm based on the visual large model also includes the following steps:

[0085] Step S240, based on the tag data set and the preset second correspondence list, determine the business product corresponding to the tag data set in the second correspondence list, and mark the determined business product as the target business product of the device; the second correspondence list has a mapping relationship between business products and sub-tags.

[0086] Specifically, step S240 may be considered to be executed not after S230; the second correspondence list may include a mapping relationship between business products and tags, as shown in Table 2 (the mapping relationship shown in Table 2 is for illustration only and does not limit the above mapping relationship).

[0087] Table 2

[0088]

[0089] The tags in the tag data set are matched with the second correspondence list, and the business product corresponding to the tag data set in the second correspondence list is determined based on the matching result; and the determined business product is marked as the target business product of the device. For each device, it has user attributes and can know which user it belongs to.

[0090] For example, let's take user A's device as an example. It's determined that the tags in user A's device tag data set are {Tag 1, Tag 2, Tag 3}. User A's confirmed service product is {Fire Recognition}. By matching user A's tags with the second correspondence list, we find that user A's current service products are {Fire Recognition, Intrusion Recognition, Specific Action Recognition (Smoking, etc.)}. User A's target service products are then updated to {Fire Recognition, Intrusion Recognition, Specific Action Recognition}. At this point, user A can be considered a potential customer for fire recognition, intrusion recognition, and specific action recognition. Potential customers refer to potential customer groups.

[0091] Through this embodiment, user potential customer attribute label processing can be performed in association, the device usage environment can be combined with the user's potential application products, and the user's potential needs can be converged into clear and specific product capabilities.

[0092] In some embodiments, the method for issuing a target scene algorithm based on a large visual model further includes the following steps:

[0093] Promote the target business products to the users corresponding to the device.

[0094] Specifically, since the users corresponding to the marked devices are potential customers of the corresponding products, promotions can be carried out for the users corresponding to the devices based on these marks, thereby improving the conversion rate and reducing marketing costs.

[0095] It should be noted that the steps shown in the above process or the flowchart in the accompanying drawings can be executed in a computer system such as a set of computer-executable instructions, and although a logical order is shown in the flowchart, in some cases, the steps shown or described can be executed in an order different from that shown here.

[0096] In this embodiment, a device for issuing a target scene algorithm based on a large visual model is also provided. The device is used to implement the above-mentioned embodiments and preferred implementation modes, and the details that have been described will not be repeated. The terms "module", "unit", "sub-unit", etc. used below can implement a combination of software and / or hardware for predetermined functions. Although the devices described in the following embodiments are preferably implemented in software, implementation by hardware, or a combination of software and hardware, is also possible and conceivable.

[0097] Figure 6 This is a structural block diagram of the target scene algorithm delivery device based on the visual large model of this embodiment, such as Figure 6 As shown, the device includes: an acquisition module 210, a processing module 220 and a sending module 230;

[0098] The acquisition module 210 is used to acquire the current tag set and the last analyzed tag set based on a device with a built-in visual large model; the device is associated with various scene algorithms;

[0099] A processing module 220 is configured to determine a tag data set of a device based on the subtags in the current tag set and the subtags in the last analyzed tag set;

[0100] The sending module 230 is used to match the tag data set with the preset first correspondence list to obtain a matching result; and complete the sending of the target scene algorithm based on the matching result; the first correspondence list has a mapping relationship between the scene algorithm and the tag.

[0101] Through the above-mentioned device, the problem of limited algorithm usage scenarios in the terminal in the related technology is solved. The corresponding scene algorithm is dynamically matched according to the label, so that different scene algorithms can be issued for different scene algorithms. It can effectively prevent the scene algorithm from being interfered with by environmental factors such as weather and light, improve the accuracy of the scene algorithm issuance under scene conversion, and reduce the recall rate.

[0102] In some embodiments, the acquisition module 210 is further configured to capture a video stream based on a device with a built-in visual model, and extract picture frames from the video stream at preset time intervals;

[0103] The picture frame is subjected to the picture frame label analysis of the visual large model, and the current label set of the picture frame is output in the form of a set.

[0104] In some embodiments, the processing module 220 is further configured to compare the subtags in the current tag set with the subtags in the last analyzed tag set to determine whether there are different subtags.

[0105] If there are different sub-tags, reset the sub-tags in the last analyzed tag set to the sub-tags in this tag set;

[0106] If there is no different sub-tag, the device's tag data set is updated based on the current tag set.

[0107] In some embodiments, the sending module 230 is further configured to determine a scenario algorithm corresponding to each sub-tag based on the tag data set and a preset first correspondence list;

[0108] If multiple subtags correspond to one scene algorithm, the corresponding scene algorithm is determined as the target scene algorithm that the device should use;

[0109] If multiple subtags correspond to multiple scene algorithms, the score of each corresponding scene algorithm is calculated according to the preset scoring formula, and the scene algorithm with the highest score is determined as the target scene algorithm that the device should use;

[0110] If multiple subtags do not correspond to scene algorithms, the universal scene algorithm is determined as the target scene algorithm that the device should use.

[0111] The recommendation module score formula is weight coefficient k1×label 1+weight coefficient k2×label 2+…weight coefficient kn×label n.

[0112] In some of the embodiments, the target scene algorithm distribution device based on the visual large model further includes a promotion module;

[0113] The promotion module is used to determine the business product corresponding to the tag data set in the second correspondence list based on the tag data set and the preset second correspondence list, and mark the determined business product as the target business product of the device; the second correspondence list has a mapping relationship between business products and sub-tags.

[0114] In some embodiments, the promotion module is further configured to promote the target business product to the user corresponding to the device.

[0115] It should be noted that the above modules can be functional modules or program modules, and can be implemented through software or hardware. For modules implemented through hardware, the above modules can be located in the same processor; or the above modules can be located in different processors in any combination.

[0116] This embodiment further provides a computer device, including a memory and a processor, wherein the memory stores a computer program, and the processor is configured to run the computer program to execute the steps in any one of the above method embodiments.

[0117] Optionally, the computer device may further include a transmission device and an input / output device, wherein the transmission device is connected to the processor, and the input / output device is connected to the processor.

[0118] Optionally, in this embodiment, the processor may be configured to execute the following steps through a computer program:

[0119] S1, based on a device with a built-in large visual model, obtains the current tag set and the last analyzed tag set; the device is associated with various scene algorithms;

[0120] S2, determining the device's tag data set based on the sub-tags in the current tag set and the sub-tags in the last analyzed tag set;

[0121] S3, matching the tag data set with a preset first correspondence list to obtain a matching result; and completing the distribution of the target scene algorithm according to the matching result; the first correspondence list has a mapping relationship between the scene algorithm and the tag.

[0122] It should be noted that, for specific examples in this embodiment, reference may be made to the examples described in the above embodiments and optional implementation modes, and will not be repeated in this embodiment.

[0123] In addition, in conjunction with the method for delivering a target scene algorithm based on a large visual model provided in the above embodiments, a storage medium may also be provided in this embodiment for implementation. The storage medium stores a computer program; when executed by a processor, the computer program implements any of the methods for delivering a target scene algorithm based on a large visual model provided in the above embodiments.

[0124] It should be noted that the information and data involved in this application are all authorized by the user or fully authorized by all parties and will be used legally.

[0125] It should be understood that the specific embodiments described herein are only used to explain this application and are not used to limit it. Based on the embodiments provided in this application, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of this application.

[0126] Obviously, the accompanying drawings are merely examples or embodiments of the present application. A person skilled in the art can also apply the present application to other similar situations based on these drawings without inventive effort. Furthermore, it is understandable that, although the work involved in this development process may be complex and lengthy, certain design, manufacturing, or production changes based on the technical content disclosed in this application are merely routine technical means for a person skilled in the art and should not be considered to constitute a deficiency in the disclosure of the present application.

[0127] The term "embodiment" as used in this application refers to specific features, structures, or characteristics described in conjunction with the embodiment that can be included in at least one embodiment of this application. The appearance of this phrase in various places in the specification does not necessarily mean that the embodiment is the same, nor does it mean that it is mutually exclusive with other embodiments and is independent or optional. It is understood, either explicitly or implicitly, by those skilled in the art that the embodiments described in this application can be combined with other embodiments when there is no conflict.

[0128] The above-described embodiments merely represent several implementation methods of the present application. While the descriptions are relatively specific and detailed, they should not be construed as limiting the scope of patent protection. It should be noted that a person of ordinary skill in the art may make various modifications and improvements without departing from the spirit of the present application, and these modifications and improvements fall within the scope of protection of the present application. Therefore, the scope of protection of the present application shall be determined by the appended claims.

Claims

1. A method for issuing a target scene algorithm based on a large visual model, characterized in that: include: Based on a device with a built-in large visual model, the current tag set and the last analyzed tag set are obtained; the device is associated with various scene algorithms; Determining a tag data set for the device based on the subtags in the current tag set and the subtags in the last analyzed tag set; Matching the tag data set with a preset first correspondence list to obtain a matching result; The target scene algorithm is issued according to the matching result; the first correspondence list contains a mapping relationship between the scene algorithm and the sub-tag.

2. The method for issuing a target scene algorithm based on a large visual model according to claim 1, characterized in that: Based on the device with built-in visual large model, obtain the current label set, including: Capturing a video stream based on a device with a built-in visual model, and extracting picture frames from the video stream at preset time intervals; The picture frame is subjected to picture frame label analysis of a large visual model, and the current label set of the picture frame is output in a set form.

3. The method for issuing a target scene algorithm based on a large visual model according to claim 1, characterized in that: Determining the tag data set of the device according to the current tag set and the last analyzed tag set includes: Compare the subtags in the current tag set with the subtags in the last analyzed tag set to determine whether there are different subtags; If there are different sub-tags, the sub-tags in the last analyzed tag set are reset to the sub-tags in the current tag set; If there is no difference sub-tag, the tag data set of the device is updated based on the current tag set.

4. The method for issuing a target scene algorithm based on a visual large model according to any one of claims 1 to 3, characterized in that: Matching the tag data set with a preset first correspondence list to obtain a matching result includes: Determining a scene algorithm corresponding to each sub-tag according to the tag data set and a preset first correspondence list; If multiple sub-tags correspond to one scene algorithm, the corresponding scene algorithm is determined as the target scene algorithm to be used by the device; If multiple subtags correspond to multiple scene algorithms, the score of each corresponding scene algorithm is calculated according to a preset scoring formula, and the scene algorithm with the highest score is determined as the target scene algorithm to be used by the device; If multiple sub-tags do not correspond to scene algorithms, the universal scene algorithm is determined as the target scene algorithm that the device should use.

5. The method for issuing a target scene algorithm based on a large visual model according to claim 4, characterized in that: The score formula is weight coefficient k1×label 1+weight coefficient k2×label 2+…weight coefficient kn×label n.

6. The method for issuing a target scene algorithm based on a large visual model according to claim 4, characterized in that: The method further comprises: According to the label data set and a preset second correspondence list, determine the business product corresponding to the label data set in the second correspondence list, and mark the determined business product as the target business product of the device; the second correspondence list has a mapping relationship between the business product and the sub-label.

7. The method for issuing a target scene algorithm based on a large visual model according to claim 6, characterized in that: The method further comprises: Promotion is performed for users corresponding to the device based on the target business product.

8. A device for issuing a target scene algorithm based on a large visual model, characterized in that: include: Acquisition module, processing module and delivery module; The acquisition module is used to acquire the current tag set and the last analyzed tag set based on a device with a built-in visual large model; the device is associated with various scene algorithms; The processing module is configured to determine the tag data set of the device based on the subtags in the current tag set and the subtags in the last analyzed tag set; The issuing module is configured to perform matching based on the tag data set, the scenario algorithm, and a preset first correspondence list to obtain a matching result; The target scene algorithm is issued according to the matching result; the first correspondence list contains a mapping relationship between the scene algorithm and the sub-tag.

9. A computer device comprising a memory and a processor, characterized in that: A computer program is stored in the memory, and the processor is configured to run the computer program to execute the steps of the method for issuing a target scene algorithm based on a large visual model according to any one of claims 1 to 7.

10. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the steps of the method for issuing a target scene algorithm based on a large visual model according to any one of claims 1 to 7 are implemented.

Images