A store sofa sampling detection method and system and a storage medium

Abstract

The application relates to the technical field of furniture sales, in particular to a store sofa sampling detection method and system and a storage medium, which comprises the following steps: generating a corresponding product list according to a sales plan; obtaining plan sofa information in the product list, and obtaining a corresponding plan sofa model from a product database according to the plan sofa information; updating and adjusting the product database according to a convolutional neural network training method; obtaining an exhibited sofa image in a store, and storing the exhibited sofa image in a product picture library; identifying and comparing the plan sofa model obtained from the product database and the exhibited sofa image in the product picture library, and obtaining a corresponding comparison result; and judging whether the store sampling is qualified according to the comparison result. The application has the effect of improving the timely detection management of the enterprise end on the store sampling of sofas according to the sales plan.

Description

TECHNICAL FIELD

[0001] The application relates to the technical field of furniture sales, in particular to a store sofa display detection method and system and a storage medium. BACKGROUND

[0002] With the improvement of people's economic ability, more and more people have greater demand for household supplies, such as sofas, which tests the research and development, production and sales capabilities of more and more enterprises. More enterprises make complete and detailed sales plans for their sofas, including the display time, combination style, marketing mode and sales sequence of the sofas in stores, and the execution capability of the stores for the sales plans of the enterprises is related to whether the sales layout of the enterprises in the time period can be effectively realized.

[0003] At present, for the household industry such as sofas, when the stores display and sell the sofas according to the plans, the personnel of the enterprises generally obtain the execution information of the sales plans in the stores by actively visiting and investigating or by taking videos and photos of the displayed sofas and sending the videos and photos to the management personnel associated with the stores.

[0004] Some problems may occur by using this method. For example, when the number of stores is large, the enterprises cannot match one management personnel with one store, and one management personnel often manages multiple stores in the same area. At this time, if the management personnel actively investigates the management, a lot of time is needed, which not only increases the workload, but also cannot timely rectify and product display for the stores that do not display and sell the sofas according to the sales plans, which may affect the profits of the enterprises. The method of taking photos of the displayed sofas by the personnel of the stores and uploading the photos needs the management personnel to receive the photos and record the stores, and the management personnel and the personnel of the stores may repeatedly communicate due to the forgetting of the personnel of the stores to take photos, which wastes time. SUMMARY

[0005] In order to improve the timely detection and management effect of the enterprises on the display of the sofas in the stores according to the sales plans, the application provides a store sofa display detection method, system and storage medium.

[0006] In a first aspect, the application provides a store display detection method, which adopts the following technical solution:

[0007] A store sofa display detection method, characterized in that it comprises:

[0008] generating a corresponding product list according to a sales plan;

[0009] obtaining a planned sofa model corresponding to the planned sofa information from a product database;

[0010] updating and adjusting the product database according to a convolutional neural network training method;

[0011] obtaining an exhibited sofa image in a store and storing the exhibited sofa image in a product picture library;

[0012] identifying and comparing the planned sofa model obtained from the product database with the exhibited sofa image in the product picture library, and obtaining a corresponding comparison result;

[0013] determining whether the store appearance is qualified according to the comparison result.

[0014] In some embodiments, the planned sofa information includes a sofa single product model and a sofa combination model, and obtaining a planned sofa model corresponding to the planned sofa information from a product database includes:

[0015] matching the sofa single product model to a planned sofa single product model of the same single product model in the product database, the planned sofa single product model including a 3D rendering model of different forms under light rendering in at least four directions;

[0016] matching the sofa combination model to a planned sofa combination model of the same combination model in the product database, the planned sofa combination model including a 3D rendering model of different forms under light rendering in at least four directions.

[0017] In some embodiments, updating and adjusting the product database according to a convolutional neural network training method includes the following steps:

[0018] generating a corresponding convolutional neural network model based on the product database;

[0019] obtaining a training sofa image and labeling the training sofa image as a product model corresponding to the training sofa image;

[0020] inputting the training sofa image into the convolutional neural network model to obtain a training result;

[0021] determining whether the training result meets expectations;

[0022] if the training result meets expectations, identifying and comparing the product database with the exhibited sofa image in the product picture library;

[0023] If the training result does not meet the expectation, parameters of the convolutional neural network model are adjusted according to the training result and a product model corresponding to the training sofa image until the training result meets the expectation, and a new product database is generated according to the adjusted convolutional neural network model.

[0024] In some embodiments, the exposed sofa image in the store is obtained, including the following steps:

[0025] The product images in multiple angles are obtained by the plurality of camera modules in the store at a preset time;

[0026] It is judged whether there is an invalid image in the product image, and the invalid image includes an occluded image, a blurred image, and a high-exposure image;

[0027] If there is, the invalid image is filtered, and the remaining image after filtering is used as a preliminary screening image;

[0028] It is judged whether there is a foreign object image in the preliminary screening image, and the foreign object image represents an image of other products other than sofas;

[0029] If there is, the foreign object image is filtered, and the remaining image after filtering is used as an exposed sofa image.

[0030] In some embodiments, the plan sofa model obtained in the product database is identified and compared with the exposed sofa image in the product picture library, and a comparison result is obtained, including the following steps:

[0031] The exposed sofa image in the product picture library is input into the product database, and it is judged whether the plan sofa model matching the exposed sofa image can be identified in the product database;

[0032] If it can be identified, the comparison result is successful;

[0033] If it cannot be identified, the comparison result is failed, and the comparison result further includes a failure reason, and the failure reason includes not meeting the sales plan, single product model not meeting, and product combination not meeting.

[0034] In some embodiments, the sales plan includes a trial sales plan and a promotion plan, it is judged whether the plan sofa model associated with the trial sales plan matching the exposed sofa image can be identified in the product database, if it cannot be identified, the comparison result is failed, and the failure reason is not meeting the sales plan; it is judged whether the plan sofa model associated with the promotion plan matching the exposed sofa image can be identified in the product database, if it cannot be identified, the comparison result is failed, and the failure reason is not meeting the sales plan.

[0035] In some embodiments, when a matching sofa model in the product database is identified, an environment recognition method is further included, which comprises:

[0036] An environmental factor in the displayed sofa image is obtained, including the exhibition area background color, the exhibition area signboard, the exhibition area articles, etc.

[0037] The exhibition area style is analyzed according to the environmental factor, including the bedroom style, the living room style, the dining room style.

[0038] It is determined whether the exhibition area style is the dining room style, and if so, the comparison result is failed.

[0039] If the exhibition area style is not the dining room style, an appearance factor of the sofa is obtained, including the sofa color, the sofa type, the sofa material, etc.

[0040] The appearance factor of the sofa is compared with the environmental factor, and it is determined whether the appearance factor of the sofa matches the environmental factor, and if so, the comparison result is successful, and if not, the comparison result is failed.

[0041] In some embodiments, the store display is determined to be qualified according to the comparison result, specifically comprising the following steps:

[0042] If the comparison result is successful, the store display is qualified.

[0043] If the comparison result is failed, the store display is unqualified, and corresponding display abnormal information and display supervision information are generated, the display abnormal information and the comparison result are sent to the store, and the display supervision information and the comparison result are sent to the superior management layer associated with the store.

[0044] In a second aspect, the application provides a store display detection system, which adopts the following technical scheme:

[0045] A store sofa display detection system, characterized in that it comprises a cloud and a camera module, wherein

[0046] The camera module is used to take pictures of the displayed sofa in the store to obtain a displayed sofa image.

[0047] The cloud is used to generate a corresponding product list according to a sales plan, obtain planned sofa information in the product list, obtain a corresponding planned sofa model from a product database according to the planned sofa information, update and adjust the product database according to a convolutional neural network training method, obtain an exhibited sofa image and store the exhibited sofa image in a product picture library, identify and compare the planned sofa model obtained from the product database with the exhibited sofa image in the product picture library, and obtain a corresponding comparison result, and determine whether the store appearance is qualified according to the comparison result.

[0048] In a third aspect, the present application provides a computer readable storage medium, which adopts the following technical solution:

[0049] A computer readable storage medium, which stores a computer program, and the computer program is executed by a processor to implement the store sofa appearance detection method.

[0050] In summary, the present application has at least the following beneficial technical effects:

[0051] When the enterprise side formulates a sales plan for a period of time, a product list is made, and a corresponding planned sofa model is extracted from a product database according to the product list, and exhibited sofa images photographed by a camera module in each store are received, and the exhibited sofa images are compared and identified with the planned sofa model through a neural network model algorithm. If the corresponding planned sofa model can be matched, it means that the store appearance is qualified. If no matching is found, it means that the store appearance is unqualified, and the store and the superior management layer associated with the store are urged to perform the sofa appearance in the marketing plan, so as to improve the timely detection and management effect of the enterprise side on the sofa appearance according to the sales plan. BRIEF DESCRIPTION OF DRAWINGS

[0052] Figure 1 It is a whole process schematic diagram of the store sofa appearance detection method in the embodiment of the present application;

[0053] Figure 2 It is a flowchart of updating and adjusting the product database using a convolutional neural network method in the embodiment of the present application;

[0054] Figure 3 It is a flowchart of determining whether the store appearance is qualified in the embodiment of the present application;

[0055] Figure 4 It is an example schematic diagram of the store sofa appearance detection method in the embodiment of the present application;

[0056] Figure 5 It is a module schematic diagram of the store sofa appearance detection system in the embodiment of the present application. DETAILED DESCRIPTION

[0057] In order to more clearly understand the objects, technical solutions and advantages of the present application, the present application will be described and explained in detail below in connection with the drawings and embodiments. However, it should be understood by those of ordinary skill in the art that the present application can be implemented without these details. In some cases, in order to avoid unnecessary description and make aspects of the present application obscure, well-known methods, procedures, systems, components and / or circuits that have been described at a higher level will not be described in detail. It is obvious for those of ordinary skill in the art that various changes can be made to the embodiments disclosed in the present application, and the general principles defined in the present application can be applied to other embodiments and application scenarios without departing from the principles and scope of the present application. Therefore, the present application is not limited to the embodiments shown, but conforms to the broadest scope of the claimed range.

[0058] Unless otherwise defined, technical or scientific terms used in the present application should have the general meaning understood by those of ordinary skill in the art to which the present application belongs. The terms used in the present application are only for the purpose of describing specific embodiments, and are not intended to limit the present application. As used in the present application, "a", "an", "one", "the", "these", and similar words do not represent a quantitative limitation, but can be singular or plural. In the present application, the terms "include", "contain", "have" and any variants thereof are intended to cover non-exclusive inclusion; for example, processes, methods and systems, products or devices containing a series of steps or modules (units) are not limited to the listed steps or modules (units), but can include steps or modules (units) not listed, or can include other steps or modules (units) inherent to the processes, methods, products or devices.

[0059] In the present application, "a plurality of" means two or more. Generally, the character " / " represents that the objects associated before and after are in an "or" relationship. In the present application, the terms "first", "second", "third", etc. are only used to distinguish similar objects, and do not represent a specific order for the objects.

[0060] The terms "system", "engine", "unit", "module", and / or "block" as referred to herein are a method for differentiating different components, elements, parts, pieces, assemblies, or functions at different levels. These terms can be replaced by other expressions that can achieve the same purpose. Generally, the "module", "unit", or "block" referred to herein refers to a set of logic embodied in hardware or software instructions. The "module", "unit", or "block" described herein can be implemented as software and / or hardware, and in the case of being implemented as software, they can be stored in any type of non-transitory computer-readable storage medium or storage device.

[0061] In some embodiments, software modules / unit / blocks can be compiled and linked into executable programs. It will be appreciated that software modules can be callable from other modules / unit / blocks or from themselves, and / or can be invoked in response to detected events or interrupts. Software modules / unit / blocks configured for execution on the computing device can be stored in computer-readable storage media such as optical, digital, magnetic, or any other tangible media, or as a digital download (and can be originally stored in a compressed or installable format that needs to be installed, decompressed or decrypted prior to execution). Such software code can be stored partially or entirely in storage devices during execution, and applied in the operation of the computing device. Software instructions can be embedded in firmware, such as an EPROM. It will also be appreciated that hardware modules / unit / blocks can be included in connected logic components, such as gates and flip-flops, and / or can be included in programmable logic components, such as PLAs or processors. The modules / unit / blocks or computing device functionality described herein can be implemented as software modules / unit / blocks, which can also be stored as software modules / unit / blocks located in the memory, and can be represented in hardware or firmware. Generally, the modules / unit / blocks described herein, which can be combined with other modules / unit / blocks, or can be divided into sub-modules / sub-units / sub-blocks, although they are physically organized or stored as such. The description can apply to systems, engines, or parts thereof.

[0062] It will be understood that when a unit, engine, module, or block is referred to as being "on" or "connected" or "coupled to" another unit, engine, module, or block, it can be directly on, connected, or coupled to the other unit, engine, module, or block, or intervening units, engines, modules, or blocks can be present, unless the context clearly indicates otherwise. In this application, the term "and / or" can include any one or more of the listed items or combinations thereof.

[0063] The following detailed description of the application is made in connection with the accompanying drawings described below. Figures 1-5 The application is described in further detail.

[0064] The embodiment of the application discloses a store sofa sampling detection method.

[0065] As shown in the figure, a store sofa sampling detection method comprises: Figure 1

[0066] S100, generating a corresponding product list according to a sales plan.

[0067] The sales plan is a sales plan formulated by the headquarters, which is generally a large cycle of one year, and the sales target, sales product, sampling plan of a small cycle in the large cycle are formulated, and the research and development and production plan are determined according to the sales plan. The product list is a list directory obtained according to the sales plan, which specifies in detail the sofa style, model, time, sampling area store and the like in each time period that need to be sampled, tried and sold.

[0068] In actual use, the research and development and production efficiency may be slower or faster than the specified target, or the sales of a product may be hot, so the sales plan can be updated and modified, and when the sales plan changes, the corresponding product list is also updated and modified.

[0069] S200, obtaining plan sofa information in the product list, and obtaining a corresponding plan sofa model from a product database according to the plan sofa information.

[0070] The plan sofa information comprises a single product model and a combination model of the sofa. The single product model refers to the model of a single sofa, which includes a single sofa, a multi-person sofa that cannot be combined and split, a lazy sofa, a children's sofa and the like, and has singularity, that is, a sofa is used as a whole. The combination model refers to the model of a combination sofa, which generally refers to a sofa combined by a plurality of single sofas, a double sofa, a corner sofa, a single long sofa and an armrest sofa, and a plurality of combined sofas are used as a whole.

[0071] In S200, the corresponding plan sofa model is obtained from the product database according to the plan sofa information, which specifically comprises the following steps:

[0072] S210, matching the single product model of the sofa to the plan sofa single product model of the same single product model in the product database.

[0073] S220, matching the combination model of the sofa to the plan sofa combination model of the same combination model in the product database.

[0074] The plan sofa single product model comprises a 3D rendering model in different forms under light rendering in at least four directions, and the plan sofa combination model comprises a 3D rendering model in different forms under light rendering in at least four directions.

[0075] ​The product database in a general large enterprise stores all past sofa SKUs of all models, and the number can be as high as tens of thousands or even more. In such a large product database, we only need to find the sofa model in the product list generated by the marketing plan, because the store generally does not restock the sofas launched in the past, such as some long-term bestsellers are small probability events.

[0076] The sofa is divided into single sofa and combination sofa. According to the planned sofa information in the product list, the corresponding single sofa and combination sofa models stored in the product database can be found. When creating the models of these sofas, considering that many lights are often set in the store to irradiate the displayed goods in order to create a bright environment, in order to improve the success rate and clarity of subsequent identification, when creating the 3D model, the model can be rendered by light simulation irradiation using related software. The 3D model is sequentially subjected to multi-angle light projection to simulate the scenario of the store light projection on the sofa. In the case where the store light setting position is not clear, we can perform light rendering in four directions of the 3D model. Generally, the front, back, left and right directions of the sofa 3D model can be selected.

[0077] At the same time, in some cases, a part of some combination sofa can be a single sofa, so it is necessary to mark and separate the single sofa model and the combination sofa model in the planned sofa information to avoid automatic identification to the combination sofa when identifying the single sofa because the single sofa exists in the combination sofa.

[0078] S300, updating and adjusting the product database according to the convolutional neural network training method.

[0079] As shown in Figure 2 Because a large enterprise launches a large number of new products in each period, and the number of sofa models in the product database is also large, after decades of research and sales, it is relatively difficult to innovate the appearance and combination form of sofas. There can be improvement directions in the material, comfort, ergonomics, etc. of the sofa, but the improvement space for the appearance and combination of the sofa is small. This results in a high similarity of the appearance of the sofa at present. It is very difficult to use simple image recognition to determine whether the sofa displayed in the store is the sofa to be displayed in the sales plan, and the accuracy is low. In order to solve this problem, the convolutional neural network image recognition method is used in the embodiment of the application. Not only the product style image recognition of the sofa category is solved, but also the problem of high product similarity and easy identification error of the sofa full category is solved.

[0080] Specifically includes the following steps:

[0081] S310, generating a corresponding convolutional neural network model based on the product database.

[0082] S320, obtaining a training sofa image and labeling the training sofa image as a product model corresponding to the training sofa image.

[0083] S330, inputting the training sofa image into the convolutional neural network model to obtain a training result.

[0084] S340, determining whether the training result meets the expectation.

[0085] S350, if the training result meets the expectation, using the product database and the displayed sofa image in the product picture library for recognition comparison.

[0086] S360, if the training result does not meet the expectation, adjusting the parameters of the convolutional neural network model according to the training result and the product model corresponding to the training sofa image until the training result meets the expectation, and generating a new product database based on the adjusted convolutional neural network model.

[0087] The convolutional neural network is based on a convolutional neural network. The convolutional neural network (CNN) is a kind of feedforward neural network containing convolution calculation and having a deep structure, and is one of the representative algorithms of deep learning. The convolutional neural network includes a feature extractor composed of a convolution layer and a subsampling layer. In the convolution layer of the convolutional neural network, a neuron is connected only to part of the adjacent layer neurons. In a convolution layer of the CNN, a plurality of feature planes are usually included, each feature plane is composed of some rectangularly arranged neurons, and the neurons of the same feature plane share a weight vector, which is a convolution kernel. The convolution kernel is usually initialized in the form of a random decimal matrix, and the convolution kernel will learn to obtain a reasonable weight vector during the training process of the network. The direct benefit of the convolution kernel is to reduce the connections between the layers of the network, and at the same time to reduce the risk of overfitting.

[0088] The convolutional neural network mainly consists of an input layer, a convolution layer, a ReLU layer, a pooling layer and a fully connected layer. A complete convolutional neural network is formed by stacking these layers. In practical applications, the convolution layer and the ReLU layer are often collectively referred to as the convolution layer, so the convolution layer also needs to pass through the activation function after the convolution operation. Specifically, when the convolution layer and the fully connected layer perform transformation operations on the input, not only the activation function will be used, but also many functions, i.e. the weight vector W and the bias b of the neuron, while the ReLU layer and the pooling layer perform a fixed function operation. The parameters in the convolution layer and the fully connected layer are trained by gradient descent, so that the classification score calculated by the convolutional neural network and the label of each image in the training set can be consistent.

[0089] After the training result is obtained, the parameters of the convolutional neural network model are adjusted according to the training result and the labeled reading value, so that the training result reaches the expected effect that can be used.

[0090] The adjustment of the parameters of the convolutional neural network model includes adjustment of the learning rate, replacement of the optimizer, replacement of the initialization method, and adjustment of the network structure.

[0091] Now, the sofa is identified and compared using the OCR recognition technology, but because it uses the identified picture to compare with the pre-stored sample in the sample library, the image recognition accuracy of this algorithm is limited, and the applicability is low.

[0092] Using the neural network model for training can improve the recognition accuracy by adjusting the parameters, largely avoiding the problem of not being able to accurately identify the similar model sofas, and largely avoiding the misidentification of sofas with similar appearances after training.

[0093] The training result reaching the expected result means that the full value vector W and the bias b are as close to the real model as possible through continuous training and parameter change, so that the effect of the entire neural network is best.

[0094] The method is: first, give all the weight vectors W and the bias b random values, use the weight parameters generated by these random values to test the samples. The test value of the sample is , and the true value is y. Now define a loss function, which reflects the fitting degree of the model to the data. The worse the fitting is, the larger the value of the loss function should be. At the same time, we also expect that when the loss function is relatively large, its corresponding gradient should also be relatively large, so that the update variable can be updated faster. The goal is to make the test value as close to the true value y as possible, and the loss function is to make the loss value of the neural network as small as possible. The basic formula is: . By changing the parameters of the weight vector W and the bias b, the value of the loss function is minimized. By introducing the back propagation algorithm, the gradient of all parameters of the network model can be obtained using the back propagation algorithm. The parameters of the network are updated by the gradient descent algorithm. When the loss function converges to a certain extent, the training is ended, and the parameters of the trained neural network are saved.

[0095] Through continuous training and parameter adjustment, the model of the convolutional neural network approaches the most real model, and the recognition success rate continuously rises, meeting the requirement that when the sofa image on display is input into the convolutional neural network model, the correct result can be matched, and finally a new product database is generated, so that the planned sofa model matched from the product database can be extracted.

[0096] S400, obtaining the display sofa image in the store, and storing the display sofa image in the product picture library.

[0097] S410, acquire the multi-angle product images photographed by the several camera modules in the store at a preset time.

[0098] The store is provided with several camera modules, which can use commonly used high-definition cameras or panoramic cameras on the market, mainly for photographing in the store, and obtaining product images after photographing.

[0099] The preset time can be manually adjusted according to actual conditions, for example, the time point of snapshotting can be set at 10:00 am and 5:00 pm. At 10 o'clock in the morning, the store has just opened, and there are fewer customers, so the possibility of the photographed images being blocked by customers can be reduced. After 5 o'clock in the afternoon, the number of customers will increase, so the store is photographed again at 5 o'clock in the afternoon. After the image is photographed, the image is uploaded to the cloud. Whether the photographing work is completed is determined by whether the snapshot image is uploaded to the cloud. Each image is relatively stored with its upload time, photographing time and corresponding store.

[0100] S420, determine whether there is an invalid image in the product image.

[0101] S430, if there is, filter the invalid image and take the remaining image after filtering as the preliminary screening image.

[0102] Invalid images include images blocked by people, images blocked by lenses, images with blurred lenses, images with excessive exposure, and images blocked by foreign objects. In the store, there are often a large number of customers checking, trying and placing random objects on the sofa. After the camera module photographs the image, it may be blocked by people or objects, so that the image cannot be effectively identified. Therefore, these invalid images need to be filtered to obtain effective images that are not affected.

[0103] S440, determine whether there is a foreign object image in the preliminary screening image.

[0104] S450, if there is, filter the foreign object image and take the remaining image after filtering as the displayed sofa image.

[0105] Foreign object images represent other product images that are not sofas. In order to simulate the effect after home decoration, tea tables, TV cabinets and other objects are often placed near the sofa display area. Sometimes after changing the position of the sofa, the camera will photograph the image of other products. At this time, in order to reduce the amount of calculation, these foreign object images need to be filtered, and the remaining correct displayed sofa images need to be compared and identified.

[0106] After the invalid image and foreign object image screening, some sofa display images may be filtered out due to unqualified images. In order to ensure that all sofas in the store can be compared, the image captured by a certain camera module needs to be re-shot after the image is filtered out.

[0107] S500, the planned sofa model obtained from the product database is compared and identified with the sofa display image in the product picture library, and the corresponding comparison result is obtained.

[0108] As shown in Figure 3 and Figure 4 , the following steps are included:

[0109] S510, the sofa display image in the product picture library is input into the product database, and it is judged whether the planned sofa model matched with the sofa display image can be identified in the product database.

[0110] S520, if it can be identified, the comparison result is successful.

[0111] S530, if it cannot be identified, the comparison result is failed, and the comparison result also includes the failure reason.

[0112] The sofa display image is input into the product database, and the comparison and identification are performed according to the neural network algorithm. If the sofa in the sofa display image is the sofa that needs to be displayed in the sales plan, the planned sofa model matched with the sofa display image can be identified in the product database, and the comparison result is successful at this time, that is, the sofa in the sales plan is timely displayed in the store.

[0113] If the sofa in the sofa display image is not the sofa that needs to be displayed in the marketing plan after the comparison and identification by the neural network algorithm, the planned sofa model matched with the sofa display image cannot be identified in the product database, and the comparison result is failed at this time, that is, the sofa in the sales plan is not timely displayed in the store. When the comparison result is failed, the failure reason is also included, which includes not meeting the sales plan, single product model not meeting, and product combination not meeting.

[0114] The sales plan includes a trial plan and a promotion plan. It is judged whether the planned sofa model associated with the trial plan matched with the sofa display image can be identified in the product database. If it cannot be identified, the comparison result is failed, and the failure reason is that it does not meet the sales plan. It is judged whether the planned sofa model associated with the promotion plan matched with the sofa display image can be identified in the product database. If it cannot be identified, the comparison result is failed, and the failure reason is that it does not meet the sales plan.

[0115] A new product will have a trial period before the official promotion, and the marketing plan of the product will be adjusted according to the performance of the trial period, and the corresponding promotion will be carried out subsequently. However, because the sofas exhibited during the trial period and the promotion period are different, some sofas that do not perform well in the trial period will not be sold in the promotion period. If the sofas exhibited in the promotion plan are different from those in the trial plan, it will have a greater impact on the overall sales plan, so the comparison result is failed, and the reason for failure is that it does not meet the sales plan.

[0116] In view of the single product model not meeting, the single sofa exhibited on the store does not meet the sofa model that should be exhibited in the sales plan, and the single sofa model in the product list needs to be exhibited.

[0117] In view of the product combination not meeting, the combination sofa exhibited on the store does not meet the combination sofa model in the product list, which may be a whole error or an individual component of the combination sofa.

[0118] Further, when the product database can identify the planned sofa model matched with the exhibited sofa image, it also includes an environment recognition method, which specifically includes:

[0119] S540, obtaining the environmental factors in the exhibited sofa image.

[0120] S541, analyzing the exhibition area style according to the environmental factors.

[0121] S542, judging whether the exhibition area style is a restaurant style, if so, the comparison result is failed.

[0122] S543, if the exhibition area style is not a restaurant style, obtaining the appearance factors of the sofa.

[0123] S544, comparing the appearance factors of the sofa with the environmental factors, judging whether the appearance factors of the sofa match the environmental factors, if matching, the comparison result is successful, if not matching, the comparison result is failed.

[0124] After the salesperson places the sofa in the exhibition area, in order to restore the range of home environment, some furniture supplies will be placed around the sofa to form a certain exhibition area environment, and the style between the sofa and the surrounding environment should be consistent, otherwise if the exhibition area environment around the sofa is inconsistent with the overall style of the sofa and its harmony, to a certain extent, it will reduce the purchase desire of customers.

[0125] The environmental factors include the background color of the exhibition area, the signboard of the exhibition area, and the items in the exhibition area. The exhibition area style includes the bedroom style, the living room style, the dining room style, and the like. According to the environmental factors, the exhibition area style is analyzed. For example, if a television cabinet, a tea table, and a television are in an exhibition area, the exhibition area is analyzed to be the living room style according to the items in the exhibition area. If a bed, a bedside table, and a signboard with the words related to the bedroom are in an exhibition area, the exhibition area is analyzed to be the bedroom style.

[0126] According to the current general home environment, in a multi-room house, the combination sofa is generally placed in the living room, and the single sofa such as the lazy sofa and the armrest sofa is placed in the bedroom. In a one-room or loft house, the combination sofa or the single sofa is generally placed in the living room. In any house, the sofa is generally not placed in the dining room.

[0127] If it is judged that the exhibition area style in the sofa image is the dining room style after the comparison and analysis in the store, it means that the exhibition area style of the sofa is not suitable. At this time, the comparison result is failed, so as to reduce the possibility that the staff places the sofa in the dining room area due to some reasons.

[0128] If the exhibition area style is not the dining room style, it means that the sofa is in the bedroom area or the living room style. At this time, it is needed to judge whether the style of the sofa is consistent with the style of the living room or the bedroom area. The appearance factors of the sofa include the sofa color, the sofa type, and the sofa material. The sofa type includes the European style sofa, the American style sofa, and the Chinese style sofa. The sofa material includes the leather, the cloth, and the wood.

[0129] According to the information in the environmental factors, it is judged that the exhibition area is what style, and the appearance factors of the sofa are compared with the environmental factors.

[0130] For example, in the environmental factors in the A exhibition area, the background color of the exhibition area is black, the black European style tea table, and the European style television cabinet. The appearance factors of the sofa are the American style sofa with deep yellow color. This does not match the overall environment of the exhibition area, so the comparison result is failed.

[0131] In the environmental factors in the B exhibition area, the background color of the exhibition area is the wood color, including the wooden rattan chair, the bamboo tea table, and the calligraphy tools on the tea table. The appearance factors of the sofa are the Chinese style leather sofa with deep color. This matches the overall environment of the exhibition area, so the comparison result is successful.

[0132] In actual use, the result of the comparison can be modified according to the sales experience, the investigation of the customer feedback, and the market research.

[0133] S600, judging whether the store display is qualified according to the comparison result.

[0134] Specifically comprising:

[0135] S610, if the comparison result is successful, the store appearance is qualified.

[0136] S620, if the comparison result is failed, the store appearance is unqualified, and the corresponding appearance exception information and appearance supervision information are generated, the appearance exception information and the comparison result are sent to the store, and the appearance supervision information and the comparison result are sent to the superior management layer associated with the store.

[0137] When the store appearance is unqualified, the appearance exception information and the comparison result need to be sent to the store to urge the store to timely perform product appearance and rectification, and the appearance supervision information and the comparison result also need to be sent to the superior management layer associated with the store to urge the management layer to timely supervise and inspect the appearance management of the store, so that the store can keep up with the sales plan and sales rhythm of the enterprise and realize fine product operation management.

[0138] As shown in Figure 5 In another embodiment, a store sofa appearance detection system is also disclosed, comprising a cloud and a camera module, wherein the camera module is used to shoot the displayed sofa in the store to obtain a displayed sofa image.

[0139] The cloud comprises a smart analysis module and a smart platform module.

[0140] The smart analysis module is used to generate a corresponding product list according to a sales plan, obtain planned sofa information in the product list, obtain a corresponding planned sofa model from a product database according to the planned sofa information, update and adjust the product database according to a convolutional neural network training method, obtain the displayed sofa image and store it in a product picture library, identify and compare the planned sofa model obtained from the product database with the displayed sofa image in the product picture library, and obtain a corresponding comparison result, and judge whether the store appearance is qualified according to the comparison result.

[0141] The smart platform module receives the analysis results of the camera module and the smart analysis module, has functions of summarizing, counting, store management, access management, report generation, etc., and can manage the summarized results by store, region, province and city, and according to user permissions.

[0142] In another embodiment, a computer readable storage medium having a computer program stored thereon is also disclosed, and the computer program is executed by a processor to implement the above-mentioned store sofa appearance detection method.

[0143] The implementation principle is:

[0144] As shown in Figure 4As shown, when the enterprise side formulates a sales plan for a period of time, a product list is made, and the corresponding planned sofa model is extracted from the product database according to the product list, and the display sofa images photographed by the camera module in each store are received, and the display sofa images are compared and identified with the planned sofa model through a neural network model algorithm, if the corresponding planned sofa model can be matched, it means that the store sampling is qualified, if no matching is found, it means that the store sampling is unqualified, and the store and the superior management layer associated with the store are urged to sample the sofa in the marketing plan, so as to improve the timely detection and management effect of the enterprise side on the sampling of the sofa by the store according to the sales plan.

[0145] The above are preferred embodiments of the present application, and are not intended to limit the protection scope of the present application, therefore: any equivalent changes made on the basis of the structure, shape, principle of the present application should be covered within the protection scope of the present application.

Claims

1. A method for detecting a store sofa sampling, characterized by, The method comprises the following steps: generating a corresponding product list according to a sales plan; obtaining planned sofa information in the product list, and obtaining a corresponding planned sofa model from a product database according to the planned sofa information; updating and adjusting the product database according to a convolutional neural network training method; obtaining an exhibited sofa image in a store, and storing the exhibited sofa image in a product picture library; identifying and comparing the planned sofa model obtained from the product database with the exhibited sofa image in the product picture library, and obtaining a corresponding comparison result; judging whether a store appearance is qualified according to the comparison result; wherein the step of identifying and comparing the planned sofa model obtained from the product database with the exhibited sofa image in the product picture library, and obtaining a comparison result, comprises the following steps: inputting the exhibited sofa image in the product picture library into the product database, and judging whether a planned sofa model matching the exhibited sofa image can be identified in the product database; if yes, the comparison result is successful; if no, the comparison result is failed, and the comparison result further comprises a failure reason, which comprises not conforming to the sales plan, a single product model not conforming, and a product combination not conforming; the sales plan comprises a trial sales plan and a promotion plan, the step of judging whether a planned sofa model associated with the trial sales plan can be identified in the product database according to the exhibited sofa image, if no, the comparison result is failed, and the failure reason is not conforming to the sales plan; the step of judging whether a planned sofa model associated with the promotion plan can be identified in the product database according to the exhibited sofa image, if no, the comparison result is failed, and the failure reason is not conforming to the sales plan; when a planned sofa model matching the exhibited sofa image can be identified in the product database, further comprising an environment identification method, the environment identification method comprises: obtaining environmental factors in the exhibited sofa image, the environmental factors comprising an exhibition area background color, an exhibition area prompt card, and an exhibition area article; analyzing an exhibition area style according to the environmental factors, the exhibition area style comprising a bedroom style, a living room style, and a dining room style; judging whether the exhibition area style is the dining room style, if yes, the comparison result is failed; if the exhibition area style is not the dining room style, obtaining appearance factors of the sofa, the appearance factors comprising a sofa color, a sofa type, and a sofa material; comparing the appearance factors of the sofa with the environmental factors, and judging whether the appearance factors of the sofa match the environmental factors, if yes, the comparison result is successful, if no, the comparison result is failed.

2. The method of claim 1, wherein: the planned sofa information comprises a single product model and a combination model of the sofa, and the step of obtaining a corresponding planned sofa model from the product database according to the planned sofa information comprises: matching the single product model of the sofa to a planned sofa single product model of the same single product model in the product database, the planned sofa single product model comprising a 3D rendering model in different forms under light rendering in at least four directions. According to the sofa combination model matching the same combination model in the product database, the planned sofa combination model includes a 3D rendering model in different forms under light rendering in at least four directions.

3. The method of claim 1, wherein: According to the convolutional neural network training method, the product database is updated and adjusted, including the following steps: Based on the product database, a corresponding convolutional neural network model is generated; Obtain training sofa images, and label the training sofa images as product models corresponding to the training sofa images; Input the training sofa images into the convolutional neural network model to obtain a training result; Determine whether the training result meets expectations; If the training result meets expectations, use the product database to identify and compare with the displayed sofa images in the product image library; If the training result does not meet expectations, adjust the parameters of the convolutional neural network model according to the training result and the product model corresponding to the training sofa image until the training result meets expectations, and generate a new product database according to the adjusted convolutional neural network model.

4. The method of claim 1, wherein: Obtain the displayed sofa images in the store, including the following steps: Obtain the multi-angle product images captured by the camera modules in the store at a preset time; Determine whether there are invalid images in the product images, including blocked images, blurred images, and high-exposure images; If so, filter the invalid images and use the remaining filtered images as preliminary screening images; Determine whether there are foreign object images in the preliminary screening images, which represent other product images that are not sofas; If so, filter the foreign object images and use the remaining filtered images as displayed sofa images.

5. The method of claim 1, wherein: According to the comparison result, determine whether the store display is qualified, specifically including the following steps: If the comparison result is successful, the store display is qualified; If the comparison result is unsuccessful, the store display is unqualified, and corresponding display exception information and display supervision information are generated, the display exception information and the comparison result are sent to the store, and the display supervision information and the comparison result are sent to the superior management layer associated with the store.

6. A store sofa onboarding detection system, characterized by: It includes a cloud and a camera module, wherein, The camera module is used to capture the displayed sofa in the store to obtain the displayed sofa image; The cloud is used to generate a corresponding product list according to the sales plan, obtain the planned sofa information in the product list, obtain the corresponding planned sofa model from the product database according to the planned sofa information, update and adjust the product database according to the convolutional neural network training method, obtain the displayed sofa image, and store the displayed sofa image in the product image library; identify and compare the planned sofa model obtained from the product database with the displayed sofa image in the product image library, and obtain a corresponding comparison result; and determine whether the store display is qualified according to the comparison result; The comparison between the planned sofa model obtained from the product database and the displayed sofa image in the product image library and the comparison result include the following steps: inputting the displayed sofa image in the product picture library into the product database, and determining whether a planned sofa model matching the displayed sofa image can be identified in the product database; if yes, the comparison result is successful; if no, the comparison result is failed, and the comparison result further includes a failure reason, the failure reason including not meeting the sales plan, single product model not meeting, product combination not meeting; the sales plan includes a trial plan and a promotion plan, determining whether a planned sofa model associated with the trial plan matching the displayed sofa image can be identified in the product database, if no, the comparison result is failed, and the failure reason is not meeting the sales plan; determining whether a planned sofa model associated with the promotion plan matching the displayed sofa image can be identified in the product database, if no, the comparison result is failed, and the failure reason is not meeting the sales plan; when the product database can identify the planned sofa model matching the displayed sofa image, further including an environment identification method, the environment identification method including: obtaining environmental factors in the displayed sofa image, the environmental factors including an exhibition area background color, an exhibition area signboard, and an exhibition area article; analyzing an exhibition area style according to the environmental factors, the exhibition area style including a bedroom style, a living room style, and a dining room style; determining whether the exhibition area style is the dining room style, if yes, the comparison result is failed; if the exhibition area style is not the dining room style, obtaining appearance factors of the sofa, the appearance factors including a sofa color, a sofa type, and a sofa material; comparing the appearance factors of the sofa with the environmental factors, and determining whether the appearance factors of the sofa match the environmental factors, if yes, the comparison result is successful, if no, the comparison result is failed.

7. A computer-readable storage medium having stored thereon a computer program, characterized in that, the computer program is executed by the processor to realize the store sofa display detection method in any one of claims 1 to 5. the computer program is executed by the processor to realize the store sofa display detection method in any one of claims 1 to 5.

Images