CN120450589A - An artificial intelligence-based warehouse area management and textile fashion style application system - Google Patents
An artificial intelligence-based warehouse area management and textile fashion style application systemInfo
- Publication number
- CN120450589A CN120450589A CN202510548522.3A CN202510548522A CN120450589A CN 120450589 A CN120450589 A CN 120450589A CN 202510548522 A CN202510548522 A CN 202510548522A CN 120450589 A CN120450589 A CN 120450589A
- Authority
- CN
- China
- Prior art keywords
- data
- real
- time
- initial parameters
- intention
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Pending
Links
Landscapes
- Management, Administration, Business Operations System, And Electronic Commerce (AREA)
Abstract
本发明公开了一种基于人工智能的仓储区域管理与纺织品流行风格应用系统，涉及数据分析技术领域。该基于人工智能的仓储区域管理与纺织品流行风格应用系统，包括数据采集模块，用于识别纺织品的目标数据，所述目标数据包括文字数据、图像数据和文字图像关联数据，将目标数据进行动态适应性处理得到动态输入数据并输入至交互模块；交互模块，用于根据用户的输入指令对动态输入数据进行数据的确定和反馈调整；预测模型模块，用于根据确定和选取的数据进行分析并输出流行数据。解决了现有技术中数据量大时预测过程机械呆板、缺乏动态筛选机制的问题，进而导致的系统卡顿、操作流程不畅等技术缺陷。
The present invention discloses an artificial intelligence-based storage area management and textile fashion style application system, relating to the field of data analysis technology. This artificial intelligence-based storage area management and textile fashion style application system includes a data acquisition module for identifying target data for textiles, including text data, image data, and text-image associated data. The target data is dynamically and adaptively processed to obtain dynamic input data, which is then input into an interaction module; an interaction module for determining and providing feedback adjustments to the dynamic input data based on user input instructions; and a prediction model module for analyzing and outputting fashion data based on the determined and selected data. This system addresses the existing technical issues of a mechanical and rigid prediction process and a lack of dynamic screening mechanisms when large amounts of data are involved, which in turn lead to system freezes and poor operational processes.
Description
Technical Field
The invention relates to the technical field of data analysis, in particular to a warehouse area management and textile fashion style application system based on artificial intelligence.
Background
With the continuous development of intelligent and automatic technologies, the internet of things management system gradually evolves to a more efficient and intelligent direction. Large-scale internet of things and logistics management, especially in the textile industry, face the problems of complex inventory management, transportation scheduling, inventory updating and the like. The traditional internet of things management mode is low in efficiency, human errors are easy to occur, and the change of market demands cannot be responded in real time. Therefore, the intelligent system based on the large model provides an automatic solution for Ethernet management, and through artificial intelligence and data analysis, accurate inventory control, intelligent demand prediction and automatic scheduling can be realized, so that the efficiency and accuracy of Ethernet management are improved to a great extent.
Automation of internet of things (IoT) management is typically achieved in the prior art by integrating big data analytics, ioT and Artificial Intelligence (AI). First, the system will track each product in real time using RFID technology, ensuring that inventory information is updated in real time. Then, the algorithm based on the large model can analyze the application of historical sales to the fashion style of the textile, and the system can also provide customized fashion trend prediction according to the purchasing behavior and fashion trend of consumers, so as to help production and sales teams adjust product design and popularization strategies, thereby better meeting market demands.
In the prior art, in the aspect of design, the system acquires multidimensional data sources such as consumer behavior purchase, social media trend, fashion show and the like through big data analysis, captures and predicts the fashion trend in real time, thereby providing accurate direction for textile design. The intelligent internet of things inventory management system can help design team to adjust product style and style in time, can be used for intelligent internet of things management, optimizes inventory layout and replenishment strategy by using prediction results of popular elements, reduces backlog of stagnant sales commodity by predicting demands of different styles and sizes, and improves inventory turnover rate. In addition, by combining with the intelligent internet of things technology, the system can also automatically adjust the storage mode of stock materials and products.
In the prior art, through acquiring multidimensional data sources such as consumer behavior purchase, social media trend, fashion show, and the like, related information is usually captured and predicted in real time, staff performs statistical analysis on related data, data wanted by a designer are not easy to find directly due to large data size, the analysis process is complex, operability is poor, in addition, in the aspect of the analysis through the fashion trend, model prediction is directly performed through captured data, firstly, the data size of a prediction model is large, a lot of useless data or data which is not matched with user requirements are captured and counted together, a large amount of calculation resources are consumed, a large deviation between a result and expected requirements is caused, in addition, in the process of capturing predictive data, an audit mechanism is lacked, and the direction control with a user is caused, the prediction process is relatively mechanical, and is not flexible enough, if the conventional prediction model is combined with manual audit, and has the problems of blocking and unsmooth process and complicated actual operation due to the large data size and no screening mechanism.
Disclosure of Invention
Technical problem to be solved
Aiming at the defects of the prior art, the invention provides an artificial intelligence-based storage area management and textile popular style application system, which solves the problems that in the prior art, model prediction is directly carried out through captured data, firstly, the data volume of a prediction model is large, a lot of useless data or data which is not matched with the requirements of a user are captured and counted together, a large amount of calculation resources are consumed, a result has a large deviation from the expected requirements, and in addition, when the predictive data capture is carried out, an auditing mechanism is lacked, and the direction control of the user is carried out, the prediction process is relatively mechanical and inflexible, and if the conventional prediction model is combined with manual auditing, the problem of no screening mechanism is caused due to the large data volume, the process is blocked, and the actual operation is relatively complicated.
Technical proposal
The storage area management and textile fashion style application system based on the artificial intelligence comprises a data acquisition module, an interaction module, a prediction model module, a product fashion management module and a product fashion management module, wherein the data acquisition module is used for identifying target data of textiles, the target data comprises text data, image data and text image associated data, dynamic adaptive processing is carried out on the target data to obtain dynamic input data, the dynamic adaptive processing is carried out on the target data to obtain the dynamic input data, the target data of a target data source are acquired according to initial parameters, the initial parameters are used for controlling quality of the acquired image data, coordinated analysis is carried out according to data quality and performance data and real-time adjustment is carried out on the initial parameters, the performance data comprises network data and hardware performance data, coordinated analysis is used for balancing data instantaneity and data quality and obtaining first correction data used for correcting the initial parameters, the corrected initial parameters are recorded as first real-time acquisition parameters, the image data is acquired according to the first real-time parameters, the dynamic input data is obtained by grabbing the image data, the dynamic input data is carried out according to the first real-time parameters, the dynamic input data is determined and fed back and adjusted according to input instructions of users, the prediction model module is used for carrying out data determination and data feedback adjustment on the dynamic input data according to the input instructions of users, the data is carried out on the data, the data is carried out according to the determined and selected data and is used for outputting and is used for prompting quality of the data, but is not limited to be used for managing fashion data.
Further, the initial parameters include resolution and compression ratio.
Further, the data acquisition module further comprises a real-time intention correction unit, the first correction data is corrected secondarily according to the real-time intention data of the user to obtain second correction data, the initial parameters are corrected, the corrected initial parameters are recorded as second real-time grabbing parameters, the image data are grabbed according to the second real-time grabbing parameters to obtain dynamic input data, and the real-time intention data are used for measuring the intention of the user.
Further, the specific process of correcting the initial parameters is that when the first correction data is larger than the maximum value of the first preset range, the real-time performance and the comprehensive performance of the data quality meet the requirements, the resolution ratio is increased step by step and/or the compression ratio is reduced step by step until the first correction data is located in the first preset range, and when the first correction data is below the minimum value of the first preset range, the resolution ratio is reduced step by step and/or the compression ratio is increased step by step until the first correction data is located in the first preset range.
The method comprises the steps of carrying out first correction on first correction data according to user real-time intention data to obtain first correction data, carrying out intention analysis and quantification in a preset time period before a current moment according to input operation of a current task to obtain real-time intention data, carrying out weight distribution on the real-time intention data and the first correction data according to definition of intention analysis to obtain first correction data, and carrying out summation to obtain second correction data.
The initial parameters are further modified according to the second modification data, specifically, when the second modification data is larger than the maximum value of the second preset range, the real-time performance and the comprehensive performance of the data quality meet the requirements, the resolution is increased step by step and/or the compression rate is reduced step by step until the second modification data is located in the second preset range, when the second modification data is smaller than the minimum value of the second preset range and the real-time intention data is larger than the intention data threshold, the resolution is increased step by step and/or the compression rate is reduced step by step according to the degree of the real-time intention data until the real-time intention data is located below the intention data threshold, and when the second modification data is located below the minimum value of the second preset range and the real-time intention data is located below the intention data threshold, the resolution is reduced step by step and/or the compression rate is increased step by step until the second modification data is located in the second preset range.
Further, the data acquisition module further comprises the step of automatically setting initial parameters according to task types, wherein the task types comprise interactive design tasks and popular recommendation tasks, the resolution of the initial parameters corresponding to the interactive design tasks is higher than that of the initial parameters corresponding to the popular recommendation tasks, and/or the compression rate of the initial parameters corresponding to the interactive design tasks is lower than that of the initial parameters corresponding to the popular recommendation tasks.
Further, the method also comprises the step of initially setting initial parameters according to the average first real-time grabbing parameters or the average second real-time grabbing parameters of the last same type of task process when the tasks are initially started.
Further, the data determination and feedback adjustment of the dynamic input data according to the input instruction of the user is specifically:
The input instructions of the user comprise self-adaptive adjustment, manual adjustment, confirmation selection and discarding of data, when the input instruction is self-adaptive adjustment, target data are captured according to a mode of the data acquisition module, confirmation selection or discarding of the input instruction of the data are further carried out, when the input instruction is manual adjustment, automatic correction of initial parameters by the data acquisition module is suspended, manual adjustment is carried out on the initial parameters according to manually inputted adjustment amplitude, target data are captured according to the manually adjusted initial parameters, confirmation selection or discarding of the input instruction of the data is further carried out on the target data, when the input instruction is self-adaptive adjustment after manual adjustment, initial parameters of the data acquisition module after manual adjustment are initialized, and interaction is carried out according to steps of the self-adaptive adjustment instruction.
Further, the specific steps of analyzing and outputting the popular data according to the determined and selected data are that the determined and selected data are preprocessed and key characteristics are extracted, wherein the key characteristics comprise popular style and popular materials, the characteristic history sales volume, regional data and time information extracted by combining the text image associated data are taken as the input of a prediction model, the prediction model outputs label data, and the label data comprise popular style, popular materials and demand
Advantageous effects
The invention has the following beneficial effects:
(1) The application relates to an artificial intelligence based warehousing area management and textile fashion style application system, which aims to perform balanced analysis on two aspects of data quality and performance data, wherein the data quality is high, more hardware resources are inevitably consumed, network and hardware resources are poor, the data quality is inevitably influenced, and then the balance is performed, so that a mode which meets the requirements of users better is important, the application integrates the data quality and the performance data, the performance data comprises network data and hardware performance data, the method has the advantages that compared with the traditional independent network or hardware analysis, the nonlinear relation exists between the network and the hardware, and the hardware change can cause the network change, so the change is usually not single, and the obtained comprehensive data after the comprehensive analysis can provide a good quantitative basis for adjusting grabbing parameters, and the image data is grabbed according to the first real-time grabbing parameters to obtain dynamic input data; according to the data captured by the method, the data quality can be improved as much as possible on the premise of not influencing basic interactive operation, and in order to prevent excessive adjustment, the related threshold is set for limiting, and the basic interactive operation requirement can ensure that a user keeps higher efficiency and experience in the use process.
(2) According to the storage area management and textile fashion style application system based on artificial intelligence, through the real-time intention correction unit, the intention of a user can be analyzed in real time, the intention of the user is roughly judged in the analysis process, the initial parameters can be further regulated after the intention of the user is known, the regulation is more accurate, the intention analysis introduced here can directly judge whether the user desires better image quality or needs faster speed and fluency when carrying out simple flow, the system can dynamically regulate intention data in real time, the regulation process does not need to be set manually, the system can automatically recognize the intention, so that the whole interaction process is smoother and intelligent, the process is fused with the first correction data before, the initial balance standard is not lost, the experience sense and the use efficiency of the user are further improved, the data quality is not sacrificed, the computing resource is saved to a certain extent, and more people can use the system conveniently.
(3) According to the storage area management and textile fashion style application system based on artificial intelligence, different task initial selections are provided for an initial interface of the system, different tasks are selected according to initial settings, a designer usually needs to know more details when the task is designed interactively, a large amount of data needs to be counted, and a faster speed may be required for meeting the rapidness, but in the actual use process, different selections are provided, for example, the designer can also conduct rapid browsing, the requirement of high-quality images may also appear in the interactive process of the task is met, the initial settings are only aimed at the first time use or are not applicable for a long time, the initial operations are conducted on users, the sense of initial experience is facilitated to be increased, in the further use process, and the initial settings of the initial parameters are conducted according to the first corrected data or the second corrected data, namely, the memory function is memorized according to the average first real-time grabbing parameters or average second real-time grabbing parameters of the previous task process, the user operation habit of the user is kept, or the last time operation habit of the user is met, the user can be met, the condition that the user is not applicable for the initialization is met when the user needs to change the network habit of the system repeatedly, and the condition of the user is usually met.
Of course, it is not necessary for any one product to practice the invention to achieve all of the advantages set forth above at the same time.
Drawings
FIG. 1 is a block diagram of a system of the present invention;
fig. 2 is a flow chart of a warehouse area management and textile fashion style application system based on artificial intelligence.
Detailed Description
The following description of the embodiments of the present invention will be made clearly and completely with reference to the accompanying drawings, in which it is apparent that the embodiments described are only some embodiments of the present invention, but not all embodiments. All other embodiments, which can be made by those skilled in the art based on the embodiments of the invention without making any inventive effort, are intended to be within the scope of the invention.
In the description of the present invention, it should be understood that the terms "open," "upper," "lower," "thickness," "top," "middle," "length," "inner," "peripheral," and the like indicate orientation or positional relationships, merely for convenience in describing the present invention and to simplify the description, and do not indicate or imply that the components or elements referred to must have a particular orientation, be constructed and operated in a particular orientation, and thus should not be construed as limiting the present invention.
Referring to fig. 1-2, an embodiment of the present invention provides a technical solution, including a data acquisition module for identifying target data of textiles, where the target data includes text data, image data and text image related data, and the text image related data can be implemented by deep learning and natural language processing technology. For example, the text of the textile may contain information about the brand, material, style, color, use, etc., while the image represents a specific appearance of the textile, such as the appearance of apparel, the actual appearance of the color, details of the pattern, etc. The text association mode enables the system to not only identify single image features or text information, but also more accurately classify and identify images according to text description contents. For example, if the system recognizes that the text description mentions a "cotton T-shirt," it will guide the search and recognition of the corresponding blue image by the "blue" and "cotton" information in the text, finding the corresponding image that meets these characteristics. The text association data can effectively improve the precision of textile identification, enhance the marking level of the system and help to realize faster and accurate product identification.
Performing dynamic adaptive processing on the target data to obtain dynamic input data and inputting the dynamic input data to the interaction module;
The specific process of obtaining the dynamic input data by carrying out dynamic adaptive processing on the target data comprises the following steps:
the method comprises the steps of capturing target data sources such as electronic commerce platforms, social media, fashion websites and blogs, textile manufacturer and design company websites, and target data such as some open image data according to initial parameters, wherein different target data can be captured, such as new patterns and burst patterns of each platform are required to be searched for obtaining popular data, target design is carried out, such as a certain pattern or fabric is required to be obtained, searching can be carried out through keywords and relevance and can be carried out through image similarity, the system can locally store picture libraries of different patterns, colors and fabrics, elements of each platform can be searched approximately according to the image similarity, the initial parameters are used for controlling the quality of captured image data, the initial parameters comprise resolution, compression rate, chromaticity, image formats (such as JPEG, PNG, TIFF) and the like, but the parameters which are relatively important for design and prediction are not required to be adjusted in general, the image formats are not required to be adjusted for stable output and data fusion, the image formats are not required to be adjusted in general, the resolution and the compression rate are required to be adjusted through dynamic adjustment and the compression rate, capturing quality of images can be changed, capturing quality and capturing quality of images can be improved, the quality of the images can be easily achieved, the quality of the images can not be easily obtained through the dynamic adjustment and the compression rate is not required to meet the requirements, and the quality of the image capturing is not required to be easily achieved, and the quality is required to be easily and the quality is not required to be easily selected, and the quality is required to meet the quality by the quality.
The method comprises the steps of carrying out coordinated analysis according to data quality and performance data, and carrying out real-time adjustment on initial parameters, wherein the performance data comprise network data and hardware performance data, and generally, the resolution and the compression ratio of an image can be dynamically adjusted according to the current bandwidth, delay and other factors by analyzing the network data in real time. For example, when the network bandwidth is low, the system can reduce the resolution of the image or increase the compression rate, so as to reduce the data transmission amount, ensure that the image can be smoothly transmitted without causing any blocking or loading delay, and adjust the real-time performance (such as CPU, GPU load, memory usage rate, etc.) of the hardware. When the hardware load is higher, the complexity of image processing can be reduced, the resolution is reduced or the details of the image are reduced so as to avoid overload and ensure that the system keeps running smoothly, but how to coordinate these factors is the key point of the invention, and the aim of the coordination is to perform balanced analysis from two aspects of data quality and performance data, because the data quality is high, more hardware resources are inevitably consumed, the network and hardware resources are not good, the data quality is inevitably affected, and then how to perform the balancing is achieved, so that a way which meets the requirements of users better is obtained, as follows:
The method comprises the steps of performing coordinated analysis on real-time data and data quality of the balance data, obtaining first correction data for correcting initial parameters, and recording the corrected initial parameters as first real-time grabbing parameters;
The specific process of correcting the initial parameters comprises the steps of increasing resolution and/or reducing compression rate step by step when the first correction data is larger than the maximum value of a first preset range, wherein the first correction data is higher than the maximum value of the first preset range, the first preset range can indicate that the user experience sense processing critical state is poor, such as poor performance can lead to poor experience, scoring a large number of historical use cases, judging the time of the critical state, marking, removing a continuous section of data range according to the first correction data with the higher score of the critical state and the corresponding time of the compression rate, and selecting the continuous section of data range as the first preset range.
When the first correction 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 correction data is located in the first preset range.
In this case, the interactive effect is prioritized, and because the data quality is good in time, the system is hard to use due to the fact that the system is stuck, the experience is very bad, and the next work is hard to carry out.
According to the method, the captured data can be ensured to be improved as much as possible on the premise of not influencing basic interactive operation, the related threshold value is set for limiting in order to prevent excessive adjustment, and the basic interactive operation requirement can ensure that a user keeps higher efficiency and experience in the use process.
The interaction module is used for determining and feeding back and adjusting the dynamic input data according to the input instruction of the user;
the prediction model module is used for analyzing and outputting popular data according to the determined and selected data, wherein the popular data comprises but is not limited to textile style data and texture data, regional data and predicted quantity data;
And the warehouse management module is used for carrying out management prompt on warehouse products in each area according to the popular data.
The data acquisition module further comprises a real-time intention correction unit, the real-time intention correction unit is used for carrying out secondary correction on the first correction data according to the real-time intention data of the user to obtain second correction data and correcting the initial parameters, the corrected initial parameters are recorded as second real-time grabbing parameters, the image data are grabbed according to the second real-time grabbing parameters to obtain dynamic input data, and the real-time intention data are used for measuring the intention of the tendency of the user to operate. Performing secondary correction on the first correction data according to the real-time intention data of the user to obtain second correction data and correcting the initial parameters, wherein the method specifically comprises the following steps:
according to the input operation of the current task, carrying out intention analysis and quantification in a preset time period before the current moment to obtain real-time intention data;
the real-time intention data and the first correction data are subjected to weight distribution according to the definition of intention analysis and then summed to obtain second correction data;
And correcting the initial parameters according to the second correction data.
When the second correction data is larger than the maximum value of the second preset range, the real-time performance and the comprehensive performance of the data quality are met, the resolution is increased step by step and/or the compression rate is reduced step by step until the second correction data are positioned in the second preset range;
The setting process of the second preset range is analogous to the first preset range, and scoring setting can be performed through historical data, so that fine adjustment can be performed if the sensitivity of the system is not ideal in the actual use process after the setting of the second preset range or the first preset range.
When the second correction data is below the minimum value of the second preset range (when the system usually has longer delay and is stuck) and the real-time intention data is larger than the intention data threshold (when the real-time intention data is more desirous of image quality, the resolution is increased step by step and/or the compression rate is reduced step by step according to the degree of the real-time intention data until the real-time intention data is below the intention data threshold (when the real-time intention data is in pursuit of speed);
And when the second correction data is below the minimum value of the second preset range and the real-time intention data is below the intention data threshold (pursuing speed), the resolution is gradually reduced and/or the compression rate is gradually increased until the second correction data is within the second preset range.
In this embodiment, the real-time intention correction unit may analyze the intention of the user in real time, the analysis process approximately judges the intention of the user according to the operation process of the user, after knowing the intention of the user, the initial parameter may be further adjusted, the adjustment is more accurate, the intention analysis introduced here may directly judge whether the user desires better image quality or needs faster speed and fluency when doing simple flow, the system may dynamically adjust the intention data in real time, the adjustment process does not need to be set manually, the system may automatically identify the intention, so that the whole interaction process is smoother and more intelligent, and the process is fused with the previous first correction data, the initial balance standard may not be lost, further improving the experience sense and the use efficiency of the user, and also not sacrificing the data quality, the computing resource may be saved to a certain extent, and more people may use conveniently.
Specifically, the data acquisition module further comprises the step of automatically and initially setting initial parameters according to task types, wherein the task types comprise interactive design tasks and popular recommendation tasks;
The resolution of the initial parameters corresponding to the interactive design tasks is higher than that of the initial parameters corresponding to the popular recommendation tasks and/or the compression rate of the initial parameters corresponding to the interactive design tasks is lower than that of the initial parameters corresponding to the popular recommendation tasks.
The method further comprises the steps of initially setting initial parameters according to average first real-time grabbing parameters or average second real-time grabbing parameters of the task processes of the same type at the beginning of each task, wherein the average first real-time grabbing parameters are weighted according to the using time and are averaged according to the total time when the average first real-time grabbing parameters are the first real-time grabbing parameters of the task processes of the same type at the beginning. The second real-time grabbing parameters are weighted according to the using time and averaged according to the total time in the process of the last same type of task.
In this embodiment, the initial interface of the system may provide different task initial selections, different tasks are selected, the system selects different initial parameters according to the initial settings, when the task is designed interactively, the designer generally needs to know more details, when the task is recommended popular, a large amount of data needs to be counted, and a faster speed may be required to meet the rapidity, but in the actual use process, different selections may be provided, for example, the designer may also browse rapidly, the interaction process of the task is recommended popular, and the need for high quality images may also occur.
The scheme also comprises the steps that initial parameters are initially set according to the average first real-time grabbing parameters or the average second real-time grabbing parameters of the task process of the same type at the beginning of each task, namely, the initial parameters are a memory function, the last operation habit of the user is memorized, or the operation habit of the last use stage is memorized, the corresponding parameters are used for initializing the parameters of the next use, the initialization can avoid the repeated adjustment of the system each time, the habits of users of an account are generally approximately the same, and when the hardware and the network are unchanged, the user can meet the requirement of the next use only by fine adjustment, so that the experience of the user can be increased.
Specifically, the process of calculating the first correction data is:
f3＝fnet+fhw;
fcorr＝λ1×(1-f1)+λ2×f2+λ3×f3;
f 1 denotes the reaction speed component of the system, l n denotes the network delay, l p denotes the processing delay, referring to the time it takes for the system to process the data, l max denotes the maximum acceptable delay for normalizing the maximum value of the delay data, a 1、α2 denotes the weight coefficient for balancing the effect of the network delay and the processing delay on the reaction speed, and a sum of a 1、α2 is 1.
F 2 denotes a data quality component, res denotes an image resolution, c r denotes a compression ratio, the higher the compression ratio is, the lower the data quality is, c max denotes a maximum compression ratio for normalizing the compression ratio, b d denotes an image color depth, the higher the color depth is, the more the color of the image is rich, β 1、β2、β3 is a weight coefficient, which balances the effects of resolution, compression ratio and color depth on the data quality, respectively, the sum of the three weights is 1, res max denotes a maximum resolution, the normalized maximum resolution, and b max denotes a maximum color depth, and the normalized maximum color depth.
B w denotes the network bandwidth, b max the maximum network bandwidth, l n denotes the network delay.
C u represents the CPU usage, the higher the percentage (%) represents the CPU usage, the heavier the CPU load, the worse the performance, g u represents the GPU usage, the higher the percentage (%) represents the GPU usage, the heavier the GPU load, the worse the performance, m u represents the memory usage, the higher the percentage (%) represents the memory usage, and the higher the memory pressure, the worse the performance. The weight coefficients, gamma 1、γ2、γ3、γ4、γ5, are all weight coefficients and the sum of the five weights is 1, which are used to balance the effects of bandwidth, delay, CPU utilization, GPU utilization and memory utilization on device performance, respectively.
F 3 denotes a device performance component, lambda 1、λ2、λ3 denotes a weight coefficient and the sum of the three weights is 1 to balance the influence of the reaction speed, data quality, and device performance on the first correction data f corr.
The above parameters are all subjected to a unit removal process before calculation, and correlation is calculated using only the data portion.
In the present embodiment,
Specifically, the analysis process of the second correction data is:
fintent＝w1×Zavg+w2×Ctotal+w3×Savg+w4×Dtotal;
f intent denotes real-time intention data, Z avg denotes an enlargement average value, Z (T) denotes an enlargement degree of each enlarged picture, T is a proportion of the enlarged number of pictures when a user views an image, C total sets time, S avg denotes an effective scroll speed, denotes an average speed of user scrolling on a page, a ratio of an effective scroll distance to an effective scroll time, the effective scroll can be judged according to whether the user' S operation is interrupted for a long time, when the user has no operation for a long time, the time is marked as an invalid time, the time data is removed, D total denotes a total stay time, denotes a total time of the user staying on the page or the image for the time, the time is also calculated as an effective browsing time of the user, if the long time has no operation marked as an invalid time, the removal is performed, w 1、w2、w3、w4 is a weight coefficient of each corresponding parameter, and a sum of parameters is 1, and the influence degree of each parameter on the result is indicated.
The second correction data is specifically:
fcorr_new＝wintent×fintent+(1-wintent)×fcorr;
f corr_new represents second correction data, w intent represents intent definition weight coefficient, and the intent definition weight coefficient can be dynamically calculated according to the definition of user behaviors, which indicates that the intent of the user is very clear, if the user stays on a certain image or page for a long time and frequently enlarges the image, the user can be considered to clearly desire the image quality, w intent has a higher value, for example, the user can be considered to tend to pursue speed when quickly browsing the page and clicking a plurality of links, w intent has a higher value, and can also follow the change of f intent, the larger f intent is, the larger the surface user definitely desires the image quality to a certain extent, the corresponding w intent is, the specific corresponding relation can establish the mapping relation between f intent and w intent, and the corresponding w intent can be obtained according to f intent in real time.
In the present embodiment,
Specifically, the data determination and feedback adjustment for the dynamic input data according to the input instruction of the user is specifically:
The input instructions of the user comprise self-adaptive adjustment, manual adjustment, confirmation selection and discarding data, when the input instruction is self-adaptive adjustment, capturing target data according to a mode of the data acquisition module, further carrying out confirmation selection or discarding data input instruction selection on the target data, and when the input instruction is manual adjustment, suspending the automatic correction of the initial parameters by the data acquisition module, carrying out manual adjustment on the initial parameters according to the manually input adjustment range;
when the input instruction is manually adjusted and then is switched to self-adaptive adjustment, initializing initial parameters of the data acquisition module by using initial parameters after manual adjustment after the manual adjustment is finished, and further interacting according to the self-adaptive adjustment instruction.
In this embodiment, the input instructions described above include other forms for the purpose of increasing interactivity, such as a rollback adjustment instruction, allowing the user to rollback to a previous setting or automatically adjusting parameters, rather than readjusting from the current state each time. This is particularly useful in cases of mishandling or misregulation, reducing the false handling by the user in case of uncertainty, enabling a quick return to the previous state, and a one-touch optimization instruction, which allows the system to automatically select the most appropriate initial parameters according to the current environmental conditions (such as bandwidth, hardware performance, etc.). The method is suitable for users to automatically optimize the system, preview and adjust the instructions when the users do not need fine adjustment, allow the users to preview the captured data (images or characters) in real time in the process of collecting the data, and provide direct adjustment options. For example, the user may see the quality or text content of the image and decide whether to adjust the data acquisition parameters, a batch confirmation instruction, allowing the user to confirm or discard a series of data in batches without having to process each data item separately. Suitable for a scene of large-batch data acquisition, and the like.
Adaptive tuning enables the system to dynamically adjust initial parameters based on data acquisition and feedback in real time. For example, when the system detects that the network bandwidth is insufficient, the resolution or the compression ratio of the data can be automatically adjusted, the smooth grabbing process is ensured, the resources are not wasted, through the automatic adjustment, the user does not need excessive intervention, the system automatically adjusts according to the real-time change adjustment strategy, the operation efficiency is improved, the system automatically makes corresponding adjustment according to different input data and external environments (such as network conditions and equipment performances) so that the system can keep good performance in changeable environments, the user can manually adjust initial parameters, for example, the user can adjust the resolution, the compression ratio and the like of the data according to the needs, so that the grabbing and analysis of the system are ensured to meet specific requirements, and the manual adjustment can enable the user to quickly react to the specific requirements or abnormal conditions to perform personalized adjustment. The user can optimize data acquisition through fine-grained control, and can directly select or reject target data in the data acquisition process through input instructions of 'confirm selection' and 'discard data'. The data acquisition process is more transparent, a user can ensure that only relevant and high-quality data are acquired, data which do not meet the requirements or are irrelevant can be effectively prevented from being acquired, the data set on which the follow-up analysis depends is more accurate, the user is allowed to switch between manual adjustment and automatic adjustment, and the advantages of the manual adjustment and the automatic adjustment are combined. Manual adjustment may be used when flexible control is required, whereas adaptive adjustment may be relied upon in conventional situations. The user can switch back to self-adaptive adjustment after manual adjustment, and the system can continue to execute automatic adjustment under a more stable condition by using the parameters after the manual adjustment as a starting point, so that the efficiency and the precision are improved, and the initial parameters can be initialized after the switching to keep the use environment of the user as stable as possible.
Specifically, for the obtained data, the data are further used, one is to predict the flow trend, the other is to manage the warehouse of each region according to the predicted result data, the prediction model module is used for analyzing and outputting the popular data according to the determined and selected data, the picture needs to be preprocessed before being input into the deep learning model so as to better adapt to the input requirement of the model, the preprocessing comprises image size adjustment and normalization (scaling the pixel value from the range of [0,255] to the range of [0,1 ]), data enhancement, the data enhancement is an effective method for improving the generalization capability of the model, further, image feature extraction is carried out, the useful features (such as edges, textures, colors and the like) in the image are extracted, further, the label comprises popular style, popular materials and the like and is associated with other associated text data (such as historical sales, time information, the popular materials, the historical sales, the region data and the time information are correspondingly characterized, the popular style, the popular materials, the historical sales and the regional data and the time information are taken as input data, and the popular data comprise the popular data and the popular materials are not limited to the predicted materials and the popular materials and the time information; wherein the prediction model is a deep neural network model (DNN), the relevant data of the textile in the past year is collected before the system is constructed, the characteristics including style, material, region, time and the like are processed by an embedded layer (Embedding Layer), category data are converted into vectors, continuous numerical data such as historical sales and the like are directly used as input, the model framework comprises a Embedding layer for encoding the category data including style data, material and material data, region data and the like, the LSTM layer is used for processing time series data (historical sales data), the layer is used for capturing trend and seasonal modes of the historical sales, the fully connected layer (DENSE LAYER) is used for processing the characteristics output from the LSTM layer and the Embedding, and the output layer is used for outputting a prediction result as a category probability by using a Softmax activation function for style prediction. For the demand quantity prediction, a linear regression activation function is used for outputting continuous demand quantity, and a model is combined, namely for multi-input data, firstly, respective branches are combined, then, the combined characteristics are transferred to a full-connection layer, the model is further trained and verified, and data, in particular style data, material data, regional data (such as the regional distribution of market demands (such as sales conditions in different regions)), historical sales quantity data and time characteristics are input into the trained model;
The data tag data is output, wherein the tag data comprises popular styles, namely the most popular styles in a future period, and popular materials, namely the most popular materials in the future period. Demand is the expected number of demands for a particular style or material.
Specifically, the warehouse management module is used for carrying out management prompt on warehouse products in each area according to the popular data.
In this embodiment, the aim is to effectively manage and prompt the products in the warehouse according to the fashion data (such as fashion style, fashion material, demand, etc.) predicted by the model. In particular, this module is needed to assist warehouse management personnel in making inventory management decisions, such as scheduling products, restocking, reducing excess inventory, and the like.
It is noted that relational terms such as first and second, and the like are used solely to distinguish one entity or action from another entity or action without necessarily requiring or implying any actual such relationship or order between such entities or actions. Moreover, the terms "comprises," "comprising," or any other variation thereof, are intended to cover a non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements does not include only those elements but may include other elements not expressly listed or inherent to such process, method, article, or apparatus.
The preferred embodiments of the invention disclosed above are intended only to assist in the explanation of the invention. The preferred embodiments are not intended to be exhaustive or to limit the invention to the precise form disclosed. Obviously, many modifications and variations are possible in light of the above teaching. The embodiments were chosen and described in order to best explain the principles of the invention and the practical application, to thereby enable others skilled in the art to best understand and utilize the invention. The invention is limited only by the claims and the full scope and equivalents thereof.
Claims (10)
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202510548522.3A CN120450589A (en) | 2025-04-28 | 2025-04-28 | An artificial intelligence-based warehouse area management and textile fashion style application system |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202510548522.3A CN120450589A (en) | 2025-04-28 | 2025-04-28 | An artificial intelligence-based warehouse area management and textile fashion style application system |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN120450589A true CN120450589A (en) | 2025-08-08 |
Family
ID=96614940
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202510548522.3A Pending CN120450589A (en) | 2025-04-28 | 2025-04-28 | An artificial intelligence-based warehouse area management and textile fashion style application system |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN120450589A (en) |
Citations (7)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| EP1076459A2 (en) * | 1999-08-09 | 2001-02-14 | Motorola, Inc. | Data transfer system and method |
| WO2010016281A1 (en) * | 2008-08-08 | 2010-02-11 | 株式会社ニコン | Search support system, search support method, and search support program |
| KR20100031217A (en) * | 2008-09-12 | 2010-03-22 | (주)아이디스 | Apparatus and method for adjusting of compression rate of image data |
| KR20220111592A (en) * | 2021-02-02 | 2022-08-09 | 주식회사 패션에이드 | Fashion coordination style recommendation system and method by artificial intelligence |
| CN116431823A (en) * | 2023-03-23 | 2023-07-14 | 深圳市公狼科技有限公司 | Fashion analysis method, device, equipment and storage medium based on knowledge graph |
| CN119250706A (en) * | 2024-12-05 | 2025-01-03 | 湖南富丽真金家纺有限公司 | Intelligent management method and system for textile warehouse based on popular element information monitoring |
| US20250055802A1 (en) * | 2022-09-28 | 2025-02-13 | Tencent Technology (Shenzhen) Company Limited | Data transmission method and apparatus, computer-readable medium, and electronic device |
-
2025
- 2025-04-28 CN CN202510548522.3A patent/CN120450589A/en active Pending
Patent Citations (7)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| EP1076459A2 (en) * | 1999-08-09 | 2001-02-14 | Motorola, Inc. | Data transfer system and method |
| WO2010016281A1 (en) * | 2008-08-08 | 2010-02-11 | 株式会社ニコン | Search support system, search support method, and search support program |
| KR20100031217A (en) * | 2008-09-12 | 2010-03-22 | (주)아이디스 | Apparatus and method for adjusting of compression rate of image data |
| KR20220111592A (en) * | 2021-02-02 | 2022-08-09 | 주식회사 패션에이드 | Fashion coordination style recommendation system and method by artificial intelligence |
| US20250055802A1 (en) * | 2022-09-28 | 2025-02-13 | Tencent Technology (Shenzhen) Company Limited | Data transmission method and apparatus, computer-readable medium, and electronic device |
| CN116431823A (en) * | 2023-03-23 | 2023-07-14 | 深圳市公狼科技有限公司 | Fashion analysis method, device, equipment and storage medium based on knowledge graph |
| CN119250706A (en) * | 2024-12-05 | 2025-01-03 | 湖南富丽真金家纺有限公司 | Intelligent management method and system for textile warehouse based on popular element information monitoring |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| US20240346399A1 (en) | Resource scheduling method and system | |
| CN119513292B (en) | Recommendation strategy generation method based on large language model enhancement and related equipment | |
| CN115689341A (en) | Garment quality management method and system based on flexible production chain | |
| CN119226816B (en) | Data model training method and system based on display scheme | |
| CN120612259B (en) | Multi-agent automatic picture repairing system based on content analysis | |
| US20220414936A1 (en) | Multimodal color variations using learned color distributions | |
| CN119960946A (en) | Multi-task parallel processing method and system based on AI target recognition | |
| CN118350784A (en) | Intelligent Internet of things comprehensive management platform based on automatic regulation and control of big data | |
| KR20220134173A (en) | Electronic devices to predict sales volume of products and method of predicting sales volume of products using thereof | |
| US20210192597A1 (en) | System and method for user specific apparel attribute recommendation | |
| CN119127420A (en) | A method for constructing and intelligently scheduling multi-model services based on large models | |
| CN121170549B (en) | A collaborative learning method and system based on intelligent classification of edge samples and intelligent decision-making in the cloud. | |
| CN118052406A (en) | Supply chain data management method and system for home soft package | |
| Basaez et al. | The role of CRM‐SRM bolt‐ons in enterprise resource planning system: toward a customer‐oriented supply chain | |
| CN120450589A (en) | An artificial intelligence-based warehouse area management and textile fashion style application system | |
| CN111738790B (en) | Business push method and push system | |
| KR102422916B1 (en) | Method, apparatus and computer program for analyzing fashion image using artificial intelligence model of hierarchy structure | |
| US20250285171A1 (en) | Real-Time Augmented Reality Item Guide | |
| Zhu et al. | Automatic tuning of interactive perception applications | |
| CN119597855B (en) | Port geographic information service platform construction method based on GIS | |
| CN121327204A (en) | A Data Feature-Driven Intelligent Visualization Generation Method | |
| CN119598415A (en) | Data Quantitative Analysis and Factor Analysis Methods | |
| CN121836574A (en) | Warehouse-out optimization method and system based on multidimensional material attribute dynamic assignment | |
| Boffa et al. | How does Manufacturing Strategy Impact the Goals of a Firm? A Relational Framework Characterizing the Related Business Models’ Components | |
| CN121724717A (en) | Recommendation list generation method, device, equipment, medium and program product |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination |