An artificial intelligence-based warehouse area management and textile fashion style application system

Abstract

The application discloses a warehouse area management and textile popular style application system based on artificial intelligence and relates to the technical field of data analysis.The warehouse area management and textile popular style application system based on artificial intelligence comprises a data acquisition module, which is used for identifying target data of textiles, the target data including text data, image data and text-image associated data, and performing dynamic adaptability processing on the target data to obtain dynamic input data and inputting the dynamic input data into an interactive module; the interactive module is used for determining and feedback adjusting the data of the dynamic input data according to input instructions of a user; and a prediction model module is used for analyzing and outputting popular data according to the determined and selected data.The warehouse area management and textile popular style application system based on artificial intelligence solves the problems of a mechanical and stiff prediction process and a lack of a dynamic screening mechanism when the amount of data is large in the prior art, and further causes technical defects such as system lag and poor operation process.

Description

Technical Field

[0001] This invention relates to the field of data analysis technology, specifically to an artificial intelligence-based warehouse area management and textile fashion style application system. Background Technology

[0002] With the continuous development of intelligent and automated technologies, IoT management systems are gradually evolving towards greater efficiency and intelligence. Large-scale IoT and logistics management, especially in the textile industry, faces complex challenges such as inventory management, transportation scheduling, and inventory updates. Traditional IoT management methods are not only inefficient but also prone to human error and unable to respond to real-time changes in market demand. Therefore, intelligent systems based on large models provide automated solutions for IoT management. Through artificial intelligence and data analysis, they can achieve precise inventory control, intelligent demand forecasting, and automated scheduling, significantly improving the efficiency and accuracy of IoT management.

[0003] Existing technologies typically automate IoT management by integrating big data analytics, the Internet of Things (IoT), and artificial intelligence (AI). First, the system utilizes RFID technology to track each product in real time, ensuring up-to-date inventory information. Then, algorithms based on large-scale models analyze historical sales trends for textile fashion. The system can also provide customized trend forecasts based on consumer purchasing behavior and current trends, helping production and sales teams adjust product design and promotional strategies to better meet market demands.

[0004] In existing technologies, in the design field, the system utilizes big data analytics to acquire and predict fashion trends in real time by combining multi-dimensional data sources such as consumer behavior, social media trends, and fashion runways. This provides precise direction for textile design. It helps design teams adjust product styles and designs promptly. Furthermore, in smart IoT management, the system optimizes inventory layout and replenishment strategies using the predicted results of trend elements. By forecasting demand for different styles and sizes, it reduces the backlog of slow-moving goods and improves inventory turnover. In addition, combined with smart IoT technology, the system can automatically adjust the storage methods for inventory materials and products.

[0005] However, current technologies for capturing and predicting trends in real time, by combining multi-dimensional data sources such as consumer behavior, social media trends, and fashion shows, typically involve directly scraping relevant information. Staff then perform statistical analysis on this data. Due to the large volume of data, it's difficult to directly find the data designers want. The analysis process is not only cumbersome but also lacks operability. Furthermore, in terms of trend analysis, current technologies directly predict trends using scraped data. Firstly, the prediction model involves a large amount of data, including much useless or irrelevant data, which consumes significant computing resources and can lead to large deviations from expectations. Moreover, the lack of a review mechanism and user-directed control during predictive data scraping results in a mechanical and inflexible prediction process. Even when combining existing prediction models with manual review, the large data volume and lack of a filtering mechanism can cause lag, a choppy process, and cumbersome operation. Summary of the Invention

[0006] Technical problems to be solved

[0007] To address the shortcomings of existing technologies, this invention provides an AI-based warehouse area management and textile fashion style application system. This system solves the problems of existing technologies that rely on directly predicting data through modeling. Firstly, the large volume of data in the prediction model, including much useless or irrelevant data, consumes significant computing resources and leads to substantial deviations from expectations. Secondly, the lack of a review mechanism and user-directed control during predictive data collection results in a mechanical and inflexible prediction process. Even when combined with manual review, the large data volume and lack of filtering mechanisms cause lag, inefficiency, and cumbersome operation.

[0008] Technical solution

[0009] To achieve the above objectives, the present invention provides the following technical solution: an artificial intelligence-based warehouse area management and textile fashion style application system, comprising a data acquisition module for identifying target data of textiles, including text data, image data, and text-image association data; dynamically adapting the target data to obtain dynamic input data, which is then input to the interaction module; the specific process of dynamically adapting the target data to obtain dynamic input data is as follows: capturing target data from the target data source according to initial parameters, which are used to control the quality of the captured image data; coordinating and analyzing the data quality and performance data, and adjusting the initial parameters in real time; and adjusting the performance data... The system includes network data and hardware performance data; a coordination analysis is used to balance data real-time performance and data quality, and to obtain the first corrected data for correcting the initial parameters, which are denoted as the first real-time capture parameters; dynamic input data is obtained by capturing image data based on the first real-time capture parameters; an interaction module is used to determine and adjust the dynamic input data according to the user's input instructions; a prediction model module is used to analyze and output trend data based on the determined and selected data, including but not limited to the popular styles and materials of textiles and predicted demand data; and a warehouse management module is used to provide management prompts for warehouse products in various regions based on the trend data.

[0010] Furthermore, the initial parameters include resolution and compression ratio.

[0011] Furthermore, the data acquisition module also includes a real-time intent correction unit, which performs a second correction on the first correction data based on the user's real-time intent data to obtain the second correction data and corrects the initial parameters. The corrected initial parameters are recorded as the second real-time capture parameters. The image data is captured based on the second real-time capture parameters to obtain dynamic input data. The real-time intent data is used to measure the user's directional intent.

[0012] Furthermore, the specific process of correcting the initial parameters is as follows: when the first corrected data is greater than the maximum value of the first preset range, it indicates that the data real-time performance and the overall data quality performance meet the requirements. Then, the resolution is increased step by step and / or the compression ratio is decreased step by step until the first corrected data is within the first preset range. When the first corrected data is below the minimum value of the first preset range, the resolution is decreased step by step and / or the compression ratio is increased step by step until the first corrected data is within the first preset range.

[0013] Furthermore, the first corrected data is further corrected based on the user's real-time intent data to obtain the second corrected data, and the initial parameters are corrected accordingly. Specifically, the intent analysis is performed and quantified within a predetermined time period before the current moment based on the input operation of the current task to obtain real-time intent data; the real-time intent data and the first corrected data are weighted according to the clarity of the intent analysis and then summed to obtain the second corrected data; the initial parameters are corrected based on the second corrected data.

[0014] Furthermore, the initial parameters are modified according to the second modified data as follows: when the second modified data is greater than the maximum value of the second preset range, it indicates that the overall performance of data real-time performance and data quality meets the requirements, and the resolution is gradually increased and / or the compression ratio is gradually decreased until the second modified data is within the second preset range; when the second modified data is below the minimum value of the second preset range and the real-time intent data is greater than the intent data threshold, the resolution is gradually increased and / or the compression ratio is gradually decreased according to the degree of the real-time intent data until the real-time intent data is below the intent data threshold; when the second modified data is below the minimum value of the second preset range and the real-time intent data is below the intent data threshold, the resolution is gradually decreased and / or the compression ratio is gradually increased until the second modified data is within the second preset range.

[0015] Furthermore, the data acquisition module also includes automatically setting initial parameters according to the task type, which includes interaction design tasks and popular recommendation tasks; the resolution of the initial parameters corresponding to the interaction design task is higher than that of the initial parameters corresponding to the popular recommendation task and / or the compression rate of the initial parameters corresponding to the interaction design task is lower than that of the initial parameters corresponding to the popular recommendation task.

[0016] Furthermore, it also includes: setting the initial parameters based on the average first real-time capture parameters or the average second real-time capture parameters of the previous task of the same type at the beginning of each task.

[0017] Furthermore, the specific steps for determining and adjusting the dynamic input data based on the user's input instructions are as follows:

[0018] User input commands include: adaptive adjustment, manual adjustment, confirmation selection, and abandoning data. When the input command is adaptive adjustment, the target data is captured according to the method of the data acquisition module, and the user further selects whether to confirm or abandon the target data. When the input command is manual adjustment, the automatic correction of the initial parameters by the data acquisition module is paused, and the initial parameters are manually adjusted according to the manually input adjustment range. The target data is then captured based on the manually adjusted initial parameters, and the user further selects whether to confirm or abandon the target data. When the input command switches from manual adjustment to adaptive adjustment, the initial parameters after manual adjustment are used to initialize the initial parameters of the data acquisition module, and the user then interacts according to the steps of the adaptive adjustment command.

[0019] Furthermore, the specific steps for analyzing and outputting trend data based on the identified and selected data are as follows: The identified and selected data are preprocessed and key features are extracted. These key features include popular styles and popular materials. Historical sales figures, geographic data, and time information extracted from text and image association data are used as input to the prediction model. The prediction model outputs label data, which includes: popular styles, popular materials, and demand.

[0020] Beneficial effects

[0021] The present invention has the following beneficial effects:

[0022] (1) The purpose of the coordinated analysis in this application of the artificial intelligence-based warehouse area management and textile fashion style application system is to conduct a balanced analysis from the aspects of data quality and performance data. Because high data quality will inevitably consume more hardware resources, and poor network and hardware resources will inevitably affect data quality. So how to balance and obtain a more user-friendly approach is crucial. This application integrates data quality and performance data, including network data and hardware performance data, and sets up a comprehensive analysis process and corresponding results. The analysis also considers the correlation between network and hardware device performance and conducts a unified analysis. Compared with the traditional analysis of the network or hardware alone, the advantage is that there is a non-linear relationship between them. Hardware changes will also lead to network changes. Therefore, such changes are usually not isolated. After comprehensive analysis, the comprehensive data obtained can provide a good quantitative basis for adjusting the capture parameters. Dynamic input data is obtained by capturing image data according to the first real-time capture parameters. The data captured by this method can usually ensure that the data quality is improved as much as possible without affecting the basic interactive operation. In order to prevent over-adjustment, relevant thresholds are also set to limit the basic interactive operation requirements and ensure that users maintain high efficiency and experience during use.

[0023] (2) This AI-based warehouse area management and textile fashion style application system can analyze user intent in real time through a real-time intent correction unit. The analysis process is to roughly judge the user's intent based on the user's operation process. After knowing the user's intent, the initial parameters can be adjusted further. This adjustment is more precise. The intent analysis introduced here can directly determine whether the user desires better image quality or needs faster speed and smoothness for simple traffic. The system can make dynamic adjustments based on the real-time intent data. The adjustment process does not require manual settings. The system can automatically recognize the intent, making the entire interaction process smoother and more intelligent. Moreover, this process is integrated with the first correction data in the previous process without losing the initial balance standard, further improving the user experience and usage efficiency, without sacrificing data quality. To a certain extent, it can also save computing resources and facilitate more people to use it at the same time.

[0024] (3) The AI-based warehouse area management and textile fashion style application system provides different initial task selections on the initial interface. When different tasks are selected, the system selects different initial parameters according to the initial settings. When interactive design tasks are performed, designers usually need to know more details. When popular recommendation tasks are performed, a large amount of data needs to be collected. In order to meet the needs of speed, a faster speed may be required. However, in actual use, there will be different choices. For example, designers may also perform quick browsing. The interactive process of popular recommendation tasks may also require high-quality images. This initial setting is only for users who are using it for the first time or who have not used it for a long time. It helps to increase the initial experience. In the further use process, and according to the aforementioned first correction data or second correction data, real-time adjustment is made. At the beginning of each task, the initial parameters are set according to the average first real-time capture parameters or average second real-time capture parameters of the same type of task process in the previous time. This is the memory function. It remembers the user's previous operation habits or the operation habits of the last use stage and initializes the corresponding parameters for the next use. Initialization can avoid the system needing to adjust repeatedly each time. Usually, the habits of users of an account are roughly the same. When the hardware and network remain unchanged, only minor adjustments are needed to meet the needs of the next use. This can increase the user experience.

[0025] Of course, any product implementing this invention does not necessarily need to achieve all of the above advantages at the same time. Attached Figure Description

[0026] Figure 1 This is a system structure diagram of the present invention;

[0027] Figure 2This is a flowchart of the artificial intelligence-based warehouse area management and textile fashion style application system of the present invention. Detailed Implementation

[0028] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. 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 are within the scope of protection of the present invention.

[0029] In the description of this invention, it should be understood that the terms "opening", "upper", "lower", "thickness", "top", "middle", "length", "inner", "around", etc., which indicate orientation or positional relationship, are only for the convenience of describing this invention and simplifying the description, and do not indicate or imply that the components or elements referred to must have a specific orientation, or be constructed and operated in a specific orientation, and therefore should not be construed as limiting this invention.

[0030] Please see Figures 1-2 This invention provides a technical solution: an artificial intelligence-based warehouse area management and textile fashion style application system, comprising a data acquisition module for identifying target data of textiles. The target data includes text data, image data, and text-image association data. The text-image association data can be implemented using deep learning and natural language processing technologies. For example, the text on textiles may contain information such as brand, material, style, color, and purpose, while the images represent the specific appearance of the textiles, such as the appearance of the garment, the actual appearance of the color, and the details of the pattern. The text association method allows the system not only to identify single image features or text information, but also to classify and identify images more accurately based on the content of the text description. For example, if the system recognizes that the text description mentions a "cotton T-shirt," it will use the information "blue" and "cotton" in the text to guide the search and identification of corresponding blue images, finding images that match these characteristics. Text association data can effectively improve the accuracy of textile identification, enhance the system's labeling level, and help achieve faster and more accurate product identification.

[0031] The target data is dynamically and adaptively processed to obtain dynamic input data, which is then input into the interactive module.

[0032] The specific process of dynamically adapting the target data to obtain dynamic input data is as follows:

[0033] Based on initial parameters, target data is scraped from various sources such as e-commerce platforms, social media, fashion websites and blogs, textile manufacturer and design company websites, and some open image data. Different purposes may require different data. For example, to obtain trend data, it's necessary to search for new and popular items on various platforms for targeted design. To find a specific style or fabric, a search might be conducted using keywords, related searches, or image similarity. The system has a local image library with different styles, colors, and fabrics, allowing for approximate searches of elements from different platforms based on image similarity. Initial parameters control the quality of the scraped image data. Parameters include resolution, compression ratio, chroma, and image format (such as JPEG, PNG, TIFF), but chroma is usually more important for design and prediction and is not an adjustable parameter. For stable output and data fusion, the image format is also not adjusted. Usually, resolution and compression ratio are adjusted. By dynamically adjusting the resolution and compression ratio, the image capture speed and quality can be changed. Low quality is usually used for simple browsing, selection and reference of unimportant factors. In addition, when the network quality is poor and the hardware computing power is insufficient, the resolution is usually reduced or the compression ratio is increased to meet the requirements of smooth real-time interaction.

[0034] The system coordinates and analyzes data quality and performance data, and adjusts initial parameters in real time. Performance data includes network data and hardware performance data. Typically, by analyzing network data in real time, the system dynamically adjusts image resolution and compression ratio based on factors such as current bandwidth and latency. For example, when network bandwidth is low, the system can reduce image resolution or increase compression ratio to reduce data transmission and ensure smooth image transmission without stuttering or loading delays. Adjustments are also made based on real-time hardware performance (such as CPU and GPU load, memory usage, etc.). When hardware load is high, the complexity of image processing can be reduced, resolution can be lowered, or image details can be reduced to avoid overload and ensure smooth system operation. However, coordinating these factors is the focus of this invention. The purpose of coordination is to achieve a balance between data quality and performance data. High data quality inevitably consumes more hardware resources, while poor network and hardware resources will inevitably affect data quality. Therefore, how to achieve this balance to obtain a method that better meets user needs is as follows:

[0035] The coordination analysis is used to balance data real-time performance and data quality and to obtain the first corrected data for correcting the initial parameters. The corrected initial parameters are denoted as the first real-time capture parameters.

[0036] The specific process of correcting the initial parameters is as follows: When the first corrected data is greater than the maximum value of the first preset range, it indicates that the data real-time performance and overall data quality meet the requirements. Then, the resolution is increased step by step and / or the compression rate is decreased step by step until the first corrected data is within the first preset range. The first preset range can represent the critical state of user experience processing. For example, if the performance is too poor, the experience will be very bad. Alternatively, a scoring method can be used to score a large number of historical usage cases, judge the moments that are considered to be critical states, and mark them. Based on the historical moments with high critical state scores, and based on the first corrected data of the resolution and compression rate corresponding to the moments in the historical data, a small number of individual outliers are removed, and a continuous data range is selected, which is the first preset range.

[0037] When the first corrected data is below the minimum value of the first preset range, the resolution is gradually reduced and / or the compression rate is gradually increased until the first corrected data is within the first preset range.

[0038] In this situation, the effectiveness of the interaction should be given priority, because even if the data quality is good, it will be difficult to use if the system is lagging, the user experience will be very poor, and the next step of the work will be difficult to carry out.

[0039] Dynamic input data is obtained by capturing image data based on the first real-time capture parameters. Data captured in this way can usually ensure that the data quality is improved as much as possible without affecting basic interactive operations. In order to prevent over-adjustment, relevant thresholds are also set to limit the basic interactive operation requirements, ensuring that users can maintain a high level of efficiency and experience during use.

[0040] The interaction module is used to determine and adjust the dynamic input data based on the user's input instructions;

[0041] The predictive model module is used to analyze and output trend data based on the determined and selected data. Trend data includes, but is not limited to, style data and material data of textiles, regional data and predicted quantity data.

[0042] The warehouse management module is used to provide management prompts for warehouse products in various regions based on popular data.

[0043] Specifically, in addition to adjusting the parameters of the basic operations mentioned above, the system can also automatically identify requirements and make adaptive adjustments based on different real-time operations. The adaptive adjustment scheme is as follows: The data acquisition module also includes a real-time intent correction unit, which performs a second correction on the first correction data based on the user's real-time intent data to obtain the second correction data and corrects the initial parameters. The corrected initial parameters are recorded as the second real-time capture parameters. The image data is captured based on the second real-time capture parameters to obtain dynamic input data. The real-time intent data is used to measure the user's directional intent. Specifically, the second correction on the first correction data based on the user's real-time intent data to obtain the second correction data and correct the initial parameters is as follows:

[0044] Based on the input operations of the current task, perform intent analysis and quantification within a predetermined time period before the current moment to obtain real-time intent data;

[0045] The real-time intent data and the first corrected data are weighted according to the clarity of intent analysis and then summed to obtain the second corrected data;

[0046] The initial parameters are corrected based on the second correction data.

[0047] The specific steps for correcting the initial parameters based on the second correction data are as follows: when the second correction data is greater than the maximum value of the second preset range, it indicates that the data real-time performance and overall data quality meet the requirements. Then, the resolution is increased step by step and / or the compression rate is decreased step by step until the second correction data is within the second preset range.

[0048] The process of setting the second preset range is similar to that of the first preset range. It can also be set by scoring based on historical data. Of course, regardless of whether the second or first preset range is set, if you feel that the system's sensitivity performance is not up to standard during actual use, you can make fine adjustments.

[0049] When the second correction data is below the minimum value of the second preset range (at which point the system usually experiences long delays and stutters) and the real-time intent data is greater than the intent data threshold (at which point it indicates that the user desires higher image quality), the resolution is gradually increased and / or the compression rate is gradually decreased according to the level of the real-time intent data, until the real-time intent data is below the intent data threshold (at which point it indicates that the user is pursuing speed).

[0050] When the second corrected data is below the minimum value of the second preset range and the real-time intent data is below the intent data threshold (in pursuit of speed), the resolution is gradually reduced and / or the compression rate is gradually increased until the second corrected data is within the second preset range.

[0051] In this implementation, the real-time intent correction unit can analyze the user's intent in real time. The analysis process involves roughly judging the user's intent based on the user's operation process. After knowing the user's intent, the initial parameters can be further adjusted. This adjustment is more precise. The intent analysis introduced here can directly determine whether the user desires better image quality or simply needs faster speed and smoothness. The system can dynamically adjust based on the real-time intent data. The adjustment process does not require manual settings; the system can automatically recognize the intent, making the entire interaction process smoother and more intelligent. Furthermore, this process is integrated with the first correction data without losing the initial balance standard, further improving the user experience and efficiency without sacrificing data quality. To a certain extent, it can also save computing resources and facilitate simultaneous use by more people.

[0052] Specifically, the data acquisition module also includes automatically setting initial parameters according to the task type, which includes interaction design tasks and popular recommendation tasks;

[0053] The resolution of the initial parameters for the interaction design task is higher than that of the initial parameters for the popular recommendation task, and / or the compression rate of the initial parameters for the interaction design task is lower than that of the initial parameters for the popular recommendation task.

[0054] This also includes: at the start of each task, initial parameters are set based on the average first real-time capture parameter or the average second real-time capture parameter of the previous task of the same type. The average first real-time capture parameter is the first real-time capture parameter of the previous task of the same type, weighted by usage time and averaged over the total time. The second real-time capture parameter is the second real-time capture parameter of the previous task of the same type, weighted by usage time and averaged over the total time.

[0055] In this implementation scheme, the system's initial interface will provide different initial task selections. When different tasks are selected, the system will select different initial parameters based on the initial settings. When it is an interaction design task, designers usually need to know more details. When it is a popular recommendation task, a large amount of data needs to be collected. In order to meet the needs of speed, a faster speed may be required. However, in actual use, there will be different choices. For example, designers may also perform quick browsing. The interaction process of popular recommendation tasks may also require high-quality images. This initial setting is only for the first use or for users who have not used it for a long time to initialize the operation, which helps to enhance the initial experience. In the further use process, it can be adjusted in real time according to the aforementioned first correction data or second correction data.

[0056] This solution also includes initial parameter settings based on the average first real-time capture parameters or the average second real-time capture parameters of the previous similar task process at the beginning of each task. This is the memory function, which remembers the user's previous operating habits or the operating habits of the last use stage, and initializes the corresponding parameters for the next use. Initialization can avoid the system needing to adjust repeatedly each time. Usually, users of an account have roughly the same habits. When the hardware and network remain unchanged, only minor adjustments are needed to meet the requirements of the next use, which can improve the user experience.

[0057] Specifically, the process of calculating the first correction data is as follows:

[0058]

[0059] f3 = f net +f hw ;

[0060] f corr =λ1×(1-f1)+λ2×f2+λ3×f3;

[0061] f1 represents the system's reaction rate component, l n Indicates network latency, l p Processing latency refers to the time it takes for the system to process data. max α1 represents the maximum acceptable latency, used to standardize the maximum value of latency data. α2 represents the weighting coefficients, used to balance the impact of network latency and processing latency on response speed. The sum of α1 and α2 is 1.

[0062] f2 represents the data quality component, res represents the image resolution, and c r The compression ratio indicates the degree of image compression. A higher compression ratio generally means lower data quality. max Indicates the maximum compression ratio, used for normalized compression ratio, b d Represents the image color depth; the higher the color depth, the richer the colors in the image. β1, β2, and β3 are weighting coefficients, balancing the impact of resolution, compression ratio, and color depth on data quality, respectively. The sum of the three weights is 1. res max Indicates the maximum resolution, the normalized maximum resolution, b max Indicates the maximum color depth, the standardized maximum color depth.

[0063] b w Indicates network bandwidth, b max Maximum network bandwidth, l n This indicates network latency.

[0064] c uThis indicates CPU utilization, expressed as a percentage (%). A higher percentage indicates a heavier CPU load and lower performance. u This indicates GPU utilization, expressed as a percentage (%). A higher percentage indicates a heavier GPU load and lower performance. u This represents memory usage as a percentage (%), indicating how much memory is being used. A higher percentage indicates greater memory pressure and worse performance. Weighting coefficients γ1, γ2, γ3, γ4, and γ5 are weighting coefficients, and the sum of these five weights is 1. They are used to balance the impact of bandwidth, latency, CPU usage, GPU usage, and memory usage on device performance.

[0065] f3 represents the equipment performance component; λ1, λ2, and λ3 represent weighting coefficients, and the sum of these three weights is 1, used to balance the impact of reaction speed, data quality, and equipment performance on the first corrected data f. corr The impact.

[0066] All the parameters mentioned above have been deunited before calculation, and only the data portion is used to calculate the correlation.

[0067] In this implementation plan,

[0068] Specifically, the analysis process for the second corrected data is as follows:

[0069] f intent =w1×Z avg +w2×C total +w3×S avg +w4×D total ;

[0070]

[0071] f intent Z represents real-time intent data. avg Let z(t) represent the average magnification, z(t) be the magnification level of each magnified image, T be the number of magnified images, and C be the magnification value. total The preset time is used to determine the percentage of image magnification when the user views the image. (S) avg This represents the effective scrolling speed, the average speed at which a user scrolls on the page, and is the ratio of effective scrolling distance to effective scrolling time. Effective scrolling speed can be determined by whether the user's operation is interrupted for an extended period. When a user remains inactive for a long time, this time is recorded as invalid time and the data for that period is removed. totalThe total dwell time represents the total time a user spends on the page or image within this period. This time is also used to calculate the user's effective browsing time. If there is no operation for a long time, it is counted as invalid time and removed. w1, w2, w3, and w4 are the weight coefficients of each corresponding parameter, and the sum of the parameters is 1, indicating the degree of influence of each parameter on the result.

[0072] The second revised data is as follows:

[0073] f corr_new =w intent ×f intent +(1-w intent )×f corr ;

[0074] f corr_new This indicates the second correction data, w intent The intent explicitness weighting coefficient can be dynamically calculated based on the explicitness of user behavior. It indicates that the user's intent is very clear. If a user lingers on an image or page for a long time and frequently zooms in, it can be assumed that the user clearly desires higher image quality. intent A higher value, for example, if a user quickly browses a page and clicks multiple links, can be considered that the user tends to prioritize speed. intent If the value is higher, it can also follow f. intent Change, f intent A larger value for w indicates that the user clearly desires higher image quality. intent The larger the value, the more specific the correspondence can be established using f. intent with w intent The mapping relationship, based on real-time f intent The corresponding w can be obtained intent .

[0075] In this implementation plan,

[0076] Specifically, determining and adjusting the dynamic input data based on the user's input instructions involves the following steps:

[0077] User input commands include: adaptive adjustment, manual adjustment, confirmation selection, and discarding data. When the input command is adaptive adjustment, the target data is captured according to the method of the data acquisition module, and the user further selects whether to confirm or discard the target data. When the input command is manual adjustment, the automatic correction of the initial parameters by the data acquisition module is paused, and the initial parameters are manually adjusted according to the manually input adjustment range. The target data is then captured based on the manually adjusted initial parameters, and the user further selects whether to confirm or discard the target data.

[0078] When the input command is to switch from manual adjustment to adaptive adjustment, the initial parameters after manual adjustment are used to initialize the initial parameters of the data acquisition module, and then the interaction is performed according to the steps of the adaptive adjustment command.

[0079] In this implementation scheme, the input commands mentioned above include other forms, the purpose of which is to increase interactivity; for example: rollback adjustment commands allow users to roll back to previous settings or auto-adjustment parameters, instead of readjusting from the current state each time. This is particularly useful in case of accidental operation or adjustment errors, reducing user errors when uncertain and enabling quick recovery to the previous state; one-click optimization commands allow the system to automatically select the most suitable initial parameters based on current environmental conditions (such as bandwidth, hardware performance, etc.). This is suitable when users do not need fine-tuning, allowing the system to automatically optimize settings; data preview and adjustment commands allow users to preview the captured data (images or text) in real time during data acquisition and provide direct adjustment options. For example, users can see the image quality or text content and decide whether to adjust the data acquisition parameters; batch confirmation commands allow users to confirm or discard a series of data in batches without processing each data item individually. This is suitable for scenarios involving large-scale data acquisition, etc.

[0080] Adaptive adjustment enables the system to dynamically adjust initial parameters based on real-time data acquisition and feedback. For example, when the system detects insufficient network bandwidth, it automatically adjusts the data resolution or compression rate to ensure smooth data capture without wasting resources. Through automatic adjustment, users do not need to intervene excessively; the system adjusts its strategy based on real-time changes, improving operational efficiency. The system automatically makes corresponding adjustments based on different input data and external environments (such as network conditions and device performance), allowing it to maintain good performance in changing environments. Users can manually adjust initial parameters, such as data resolution and compression rate, to ensure that the system's capture and analysis meet specific needs. Dynamic adjustment allows users to react quickly and make personalized adjustments in the face of specific needs or abnormal situations. Users can optimize the data acquisition process through fine-grained control. Using "confirm selection" and "discard data" input commands, users can directly select or reject target data during the data acquisition process. This makes the data acquisition process more transparent, ensuring that only relevant and high-quality data is captured, effectively preventing the collection of unsuitable or irrelevant data, and ensuring that the dataset relied upon for subsequent analysis is more accurate. Users can switch between manual and automatic adjustment, combining the advantages of both. Manual adjustment can be used when flexible control is required, while adaptive adjustment can be relied upon under normal circumstances. Users can switch back to adaptive adjustment after manual adjustment, using the previously manually adjusted parameters as a starting point, allowing the system to continue automatic adjustment under relatively stable conditions, thereby improving efficiency and accuracy. Furthermore, initializing the initial parameters after switching can help maintain the stability of the user's operating environment as much as possible.

[0081] Specifically, the data acquired above is further utilized in two ways: first, to predict trends; and second, to manage warehouses in various regions based on the predicted data. The prediction model module analyzes and outputs trend data based on the selected data. Before inputting images into the deep learning model, preprocessing is required to better adapt to the model's input requirements. Preprocessing includes image resizing, normalization (scaling pixel values ​​from the [0,255] range to the [0,1] range), and data augmentation. Data augmentation is an effective method to improve the model's generalization ability. Then, image feature extraction is performed to extract useful features from the image (such as edges, textures, and colors). Finally, labels are constructed, including popular styles and popular materials, and associated textual data (such as historical sales and region). The system associates features with data and time information, using popular styles, popular materials, historical sales, regional data, and time information as input data. Popularity data includes, but is not limited to, popular styles and materials of textiles and predicted demand data. The prediction model is a deep neural network (DNN) model. Before system construction, relevant data on textiles from previous years are collected, including features such as style, material, region, and time. These features are processed through an embedding layer to convert categorical data into vectors, while continuous numerical data such as historical sales are directly used as input. The model architecture includes: an embedding layer for encoding categorical data such as style, material, and region; an LSTM layer for processing time-series data (historical sales data), which captures trends and seasonal patterns in historical sales; and a fully connected layer (DNN). Layer: Processes features output from the Embedding and LSTM layers. Output layer: For style prediction: Uses the Softmax activation function, outputting the prediction result as the class probability. For demand prediction: Uses the linear regression activation function, outputting continuous demand quantities. Model merging: For multiple input data, first merges the respective branches, then passes the merged features to the fully connected layer for model training and validation. Input data to the trained model includes style data, material data, geographic data (geographic distribution of market demand (e.g., sales in different regions)), historical sales data, and time features.

[0082] Output data includes tags, which include: Popular Styles: the most popular styles in the near future; Popular Materials: the most popular materials in the near future; and Demand Quantity: the estimated demand for a particular style or material.

[0083] Specifically, the warehouse management module is used to provide management prompts for warehouse products in various regions based on popularity data.

[0084] The primary objective of this implementation plan is to provide effective management guidance for products in the warehouse based on trend data predicted by the model (such as popular styles, popular materials, and demand). Specifically, this module needs to assist warehouse managers in making inventory management decisions, such as product scheduling, replenishment, and reducing excess inventory.

[0085] It should be noted that, in this document, relational terms such as "first" and "second" are used only to distinguish one entity or operation from another, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Furthermore, the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such process, method, article, or apparatus.

[0086] The preferred embodiments of the present invention disclosed above are merely illustrative of the invention. These preferred embodiments do not exhaustively describe all details, nor do they limit the invention to any specific implementation. Clearly, many modifications and variations can be made based on the content of this specification. This specification selects and specifically describes these embodiments to better explain the principles and practical applications of the invention, thereby enabling those skilled in the art to better understand and utilize the invention. The invention is limited only by the claims and their full scope and equivalents.

Claims

1. A warehouse area management and textile fashion style application system based on artificial intelligence, characterized in that, include: The data acquisition module is used to identify target data of textiles, including text data, image data and text-image association data. The target data is dynamically and adaptively processed to obtain dynamic input data and input to the interaction module. The specific process of obtaining dynamic input data by dynamically adaptively processing the target data is as follows: The target data from the target data source is captured based on initial parameters, which are used to control the quality of the captured image data. Coordinated analysis is performed based on data quality and performance data, and initial parameters are adjusted in real time. The performance data includes network data and hardware performance data. The coordination analysis is used to balance data real-time performance and data quality and to obtain the first corrected data for correcting the initial parameters. The corrected initial parameters are denoted as the first real-time capture parameters. Dynamic input data is obtained by capturing image data based on the first real-time capture parameters; The interaction module is used to determine and adjust the dynamic input data based on the user's input instructions; The predictive model module is used to analyze and output trend data based on the determined and selected data, including but not limited to the popular styles and materials of textiles and the predicted demand data. The warehouse management module is used to manage and provide prompts for warehouse products in various regions based on popularity data; The data acquisition module further includes a real-time intent correction unit, which performs a second correction on the first correction data based on the real-time intent data generated by the user's input operation on the preview data during the acquisition process, to obtain the second correction data and correct the initial parameters. The corrected initial parameters are recorded as the second real-time capture parameters. The image data is captured based on the second real-time capture parameters to obtain dynamic input data. The real-time intent data is used to measure the user's directional intent. The step of performing a second correction on the first corrected data based on the user's real-time intent data to obtain the second corrected data and correcting the initial parameters is as follows: based on the input operation of the current task, the intent is analyzed and quantified within a predetermined time period before the current moment to obtain real-time intent data. The real-time intent data and the first corrected data are weighted according to the clarity of intent analysis and then summed to obtain the second corrected data; The initial parameters are corrected based on the second correction data.

2. The warehouse area management and textile fashion style application system based on artificial intelligence according to claim 1, characterized in that: The initial parameters include resolution and compression ratio.

3. The warehouse area management and textile fashion style application system based on artificial intelligence according to claim 1, characterized in that: The specific process for correcting the initial parameters is as follows: When the first corrected data is greater than the maximum value of the first preset range, it indicates that the data real-time performance and the overall data quality performance meet the requirements. Then, the resolution is increased step by step and / or the compression rate is decreased step by step until the first corrected data is within the first preset range. When the first corrected data is below the minimum value of the first preset range, the resolution is gradually reduced and / or the compression rate is gradually increased until the first corrected data is within the first preset range.

4. The warehouse area management and textile fashion style application system based on artificial intelligence according to claim 1, characterized in that: The specific steps of correcting the initial parameters based on the second correction data are as follows: when the second correction data is greater than the maximum value of the second preset range, it indicates that the data real-time performance and the overall data quality performance meet the requirements. Then, the resolution is increased step by step and / or the compression rate is decreased step by step until the second correction data is within the second preset range. When the second correction data is below the minimum value of the second preset range and the real-time intent data is greater than the intent data threshold, the resolution is gradually increased and / or the compression rate is gradually decreased according to the degree of the real-time intent data until the real-time intent data is below the intent data threshold. When the second corrected data is below the minimum value of the second preset range and the real-time intent data is below the intent data threshold, the resolution is gradually reduced and / or the compression rate is gradually increased until the second corrected data is within the second preset range.

5. A warehouse area management and textile fashion style application system based on artificial intelligence according to claim 1 or 4, characterized in that: The data acquisition module also includes automatically setting initial parameters according to the task type, which includes interaction design tasks and popular recommendation tasks. The resolution of the initial parameters corresponding to the interaction design task is higher than that of the initial parameters corresponding to the popular recommendation task and / or the compression rate of the initial parameters corresponding to the interaction design task is lower than that of the initial parameters corresponding to the popular recommendation task.

6. The warehouse area management and textile fashion style application system based on artificial intelligence according to claim 5, characterized in that: Also includes: At the start of each task, the initial parameters are set based on the average first real-time capture parameters or the average second real-time capture parameters of the previous task of the same type.

7. The warehouse area management and textile fashion style application system based on artificial intelligence according to claim 1, characterized in that: The specific steps of determining and adjusting the dynamic input data based on the user's input instructions are as follows: User input commands include: adaptive adjustment, manual adjustment, confirmation selection, and discarding data. When the input command is adaptive adjustment, the target data is captured according to the method of the data acquisition module, and the user further selects whether to confirm or discard the target data. When the input command is manual adjustment, the automatic correction of the initial parameters by the data acquisition module is paused, and the initial parameters are manually adjusted according to the manually input adjustment range. The target data is then captured based on the manually adjusted initial parameters, and the user further selects whether to confirm or discard the target data. When the input command is to switch from manual adjustment to adaptive adjustment, the initial parameters after manual adjustment are used to initialize the initial parameters of the data acquisition module, and then the interaction is performed according to the steps of the adaptive adjustment command.

8. The warehouse area management and textile fashion style application system based on artificial intelligence according to claim 1, characterized in that: The specific steps for analyzing and outputting popular data based on the determined and selected data are as follows: The selected data is preprocessed and key features are extracted, including popular styles and popular materials. The historical sales, regional data, and time information extracted from the text and image association data are used as inputs to the prediction model. The prediction model outputs label data, which includes popular styles, popular materials, and demand.

Images