Cosmetic quality detection method and system based on big data

Abstract

The invention provides a cosmetic quality detection method and system based on big data, and relates to the technical field of cosmetic quality detection based on big data. The method comprises the following steps: determining a two-dimensional coordinate range of the whole mask, a two-dimensional coordinate range of a left eye part of the mask, a two-dimensional coordinate range of a right eye part of the mask, a two-dimensional coordinate range of a nose part of the mask and a two-dimensional coordinate range of a mouth part of the mask by using a coordinate determination model based on an image of the mask under white light illumination; constructing a plurality of nodes and a plurality of edges among the plurality of nodes; and processing the plurality of nodes and the plurality of edges among the plurality of nodes based on the graph neural network model to obtain whether the mask shape is compliant. The method can improve the compliance detection efficiency of the mask shape.

Description

Technical Field

[0001] The present invention relates to the technical field of cosmetics quality detection based on big data, and particularly relates to a method and system for cosmetics quality detection based on big data. Background Art

[0002] A facial mask is a type of cosmetic whose main application method is to be applied on the face. Facial masks usually contain various nutrients such as vitamins, amino acids, plant extracts, etc., which are used to provide additional skin nutrition and moisture, and prevent and repair skin problems. Facial masks have become an important part of users' daily skin care. However, during the use of facial masks, due to errors in the cutting of facial masks during the production process, users sometimes encounter the problem that the shape of the facial mask does not match the face, resulting in poor use effects. To solve this problem, it is sometimes necessary to conduct compliance testing on the shape of the facial mask before it leaves the factory. Traditional methods for compliance testing of facial mask shapes usually rely on manual visual inspection, which is not only inefficient but also easily affected by human factors.

[0003] Therefore, how to improve the efficiency of compliance testing for facial mask shapes is an urgent problem to be solved currently. Summary of the Invention

[0004] The main technical problem to be solved by the present invention is how to improve the efficiency of compliance testing for facial mask shapes.

[0005] According to a first aspect, the present invention provides a method for cosmetics quality detection based on big data, including: obtaining an image of a facial mask under white light illumination; using a coordinate determination model based on the image of the facial mask under white light illumination to determine the two-dimensional coordinate range of the whole facial mask, the two-dimensional coordinate range of the left eye part of the facial mask, the two-dimensional coordinate range of the right eye part of the facial mask, the two-dimensional coordinate range of the nose part of the facial mask, and the two-dimensional coordinate range of the mouth part of the facial mask; constructing multiple nodes and multiple edges between the multiple nodes, the multiple nodes including a left eye node, a right eye node, a nose node, and a mouth node, the node feature of the left eye node being the two-dimensional coordinate range of the left eye part of the facial mask, the node feature of the right eye node being the two-dimensional coordinate range of the right eye part of the facial mask, the node feature of the nose node being the two-dimensional coordinate range of the nose part of the facial mask and the two-dimensional coordinate range of the whole facial mask, and the node feature of the mouth node being the two-dimensional coordinate range of the mouth part of the facial mask; and processing the multiple nodes and the multiple edges between the multiple nodes based on a graph neural network model to obtain whether the shape of the facial mask is compliant.

[0006] Further, the method further includes: obtaining multiple images of multiple facial masks under multiple color lights; determining multiple facial masks to be verified using a lighting processing model based on the multiple images of each facial mask under multiple color lights; obtaining a stretching video of each facial mask to be verified; and determining whether there are cracks in each facial mask to be verified using a stretching video processing model based on the stretching video of each facial mask to be verified.

[0007] Further, the lighting processing model is a convolutional neural network model. The input of the lighting processing model is the multiple images of the facial mask under multiple color lights, and the output of the lighting processing model is multiple facial masks to be verified.

[0008] Further, the multiple images of the facial mask under multiple color lights include multiple images of the facial mask under red light, multiple images of the facial mask under green light, and multiple images of the facial mask under blue light.

[0009] Further, the coordinate determination model is a convolutional neural network model. The input of the coordinate determination model is the image of the facial mask under white light, and the output of the coordinate determination model is the two-dimensional coordinate range of the whole facial mask, the two-dimensional coordinate range of the left eye part of the facial mask, the two-dimensional coordinate range of the right eye part of the facial mask, the two-dimensional coordinate range of the nose part of the facial mask, and the two-dimensional coordinate range of the mouth part of the facial mask.

[0010] According to a second aspect, the present invention provides a big data-based cosmetics quality detection system, including: an acquisition module for acquiring an image of a facial mask under white light; a coordinate determination module for determining the two-dimensional coordinate range of the whole facial mask, the two-dimensional coordinate range of the left eye part of the facial mask, the two-dimensional coordinate range of the right eye part of the facial mask, the two-dimensional coordinate range of the nose part of the facial mask, and the two-dimensional coordinate range of the mouth part of the facial mask using a coordinate determination model based on the image of the facial mask under white light; a construction module for constructing multiple nodes and multiple edges between the multiple nodes. The multiple nodes include a left eye node, a right eye node, a nose node, and a mouth node. The node feature of the left eye node is the two-dimensional coordinate range of the left eye part of the facial mask, the node feature of the right eye node is the two-dimensional coordinate range of the right eye part of the facial mask, the node feature of the nose node is the two-dimensional coordinate range of the nose part of the facial mask and the two-dimensional coordinate range of the whole facial mask, and the node feature of the mouth node is the two-dimensional coordinate range of the mouth part of the facial mask; a compliance judgment module for processing the multiple nodes and the multiple edges between the multiple nodes using a graph neural network model to obtain whether the shape of the facial mask is compliant.

[0011] Further, the system is further configured to: Obtain multiple images of multiple facial masks under multiple colored lights; determine multiple facial masks to be verified based on the multiple images of each facial mask under multiple colored lights using a lighting processing model; obtain the stretching videos of each facial mask to be verified; determine whether there are cracks in each facial mask to be verified based on the stretching videos of each facial mask to be verified using a stretching video processing model.

[0012] Furthermore, the lighting processing model is a convolutional neural network model. The input of the lighting processing model is the multiple images of the facial mask under multiple colored lights, and the output of the lighting processing model is multiple facial masks to be verified.

[0013] Furthermore, the multiple images of the facial mask under multiple colored lights include multiple images of the facial mask under red light, multiple images of the facial mask under green light, and multiple images of the facial mask under blue light.

[0014] Furthermore, the coordinate determination model is a convolutional neural network model. The input of the coordinate determination model is the image of the facial mask under white light, and the output of the coordinate determination model is the two-dimensional coordinate range of the whole facial mask, the two-dimensional coordinate range of the left eye part of the facial mask, the two-dimensional coordinate range of the right eye part of the facial mask, the two-dimensional coordinate range of the nose part of the facial mask, and the two-dimensional coordinate range of the mouth part of the facial mask.

[0015] A method and system for detecting the quality of cosmetics based on big data provided by the present invention. The method includes obtaining an image of a facial mask under white light; determining the two-dimensional coordinate range of the whole facial mask, the two-dimensional coordinate range of the left eye part of the facial mask, the two-dimensional coordinate range of the right eye part of the facial mask, the two-dimensional coordinate range of the nose part of the facial mask, and the two-dimensional coordinate range of the mouth part of the facial mask based on the image of the facial mask under white light using a coordinate determination model; constructing multiple nodes and multiple edges between the multiple nodes. The multiple nodes include a left eye node, a right eye node, a nose node, and a mouth node. The node feature of the left eye node is the two-dimensional coordinate range of the left eye part of the facial mask, the node feature of the right eye node is the two-dimensional coordinate range of the right eye part of the facial mask, the node feature of the nose node is the two-dimensional coordinate range of the nose part of the facial mask and the two-dimensional coordinate range of the whole facial mask, and the node feature of the mouth node is the two-dimensional coordinate range of the mouth part of the facial mask; processing the multiple nodes and the multiple edges between the multiple nodes based on a graph neural network model to obtain whether the shape of the facial mask is compliant. This method can improve the detection efficiency of the compliance of the facial mask shape. Description of the Drawings

[0016] Figure 1 It is a schematic flow chart of a method for detecting the quality of cosmetics based on big data provided by an embodiment of the present invention; Figure 2Schematic diagram of the process for detecting the quality of facial masks by using a light - treated model provided by an embodiment of the present invention; Figure 3 Schematic diagram of a cosmetics quality detection system based on big data provided by an embodiment of the present invention. Detailed implementation manners

[0017] In an embodiment of the present invention, there is provided a Figure 1 cosmetics quality detection method based on big data as shown in the figure. The cosmetics quality detection method based on big data includes steps S1 - S4: Step S1: Obtain an image of the facial mask under white light illumination.

[0018] The facial mask is a type of cosmetics and is usually made of paper, fabric or other materials. For example, the facial mask is a patch - type facial mask.

[0019] The image of the facial mask under white light illumination is the image of the facial mask under white light illumination. White light refers to the white light in the visible spectrum.

[0020] Step S2: Based on the image of the facial mask under white light illumination, use a coordinate determination model to determine the two - dimensional coordinate range of the whole facial mask, the two - dimensional coordinate range of the left - eye part of the facial mask, the two - dimensional coordinate range of the right - eye part of the facial mask, the two - dimensional coordinate range of the nose part of the facial mask, and the two - dimensional coordinate range of the mouth part of the facial mask.

[0021] The coordinate determination model is a convolutional neural network model. The input of the coordinate determination model is the image of the facial mask under white light illumination, and the output of the coordinate determination model is the two - dimensional coordinate range of the whole facial mask, the two - dimensional coordinate range of the left - eye part of the facial mask, the two - dimensional coordinate range of the right - eye part of the facial mask, the two - dimensional coordinate range of the nose part of the facial mask, and the two - dimensional coordinate range of the mouth part of the facial mask.

[0022] The coordinate determination model can process the image of the facial mask under white light illumination to map the image of the facial mask under white light illumination into a two - dimensional coordinate system, and finally calculate the two - dimensional coordinate range of the whole facial mask, the two - dimensional coordinate range of the left - eye part of the facial mask, the two - dimensional coordinate range of the right - eye part of the facial mask, the two - dimensional coordinate range of the nose part of the facial mask, and the two - dimensional coordinate range of the mouth part of the facial mask.

[0023] Through the stacking of convolutional layers and pooling layers, the convolutional neural network can achieve the perception of local regions of the image. In this way, it can capture the detailed information in the facial mask image, such as edges, textures, etc. In the facial mask image, different parts of the facial mask have different texture and shape features, and the convolutional neural network can distinguish different parts by learning these features. The convolutional neural network can be trained by the gradient descent method.

[0024] The two-dimensional coordinate range of the whole mask refers to the two-dimensional coordinate range occupied by the image of the whole mask mapped onto a two-dimensional coordinate system. The two-dimensional coordinate range of the whole mask includes the two-dimensional coordinate ranges of the left-eye part, the right-eye part, the nose part, and the mouth part of the mask.

[0025] The two-dimensional coordinate range of the left-eye part of the mask refers to the two-dimensional coordinate range occupied by the image of the left-eye part of the mask mapped onto a two-dimensional coordinate system.

[0026] The two-dimensional coordinate range of the right-eye part of the mask refers to the two-dimensional coordinate range occupied by the image of the right-eye part of the mask mapped onto a two-dimensional coordinate system.

[0027] The two-dimensional coordinate range of the nose part of the mask refers to the two-dimensional coordinate range occupied by the image of the nose part of the mask mapped onto a two-dimensional coordinate system.

[0028] The two-dimensional coordinate range of the mouth part of the mask refers to the two-dimensional coordinate range occupied by the image of the mouth part of the mask mapped onto a two-dimensional coordinate system.

[0029] In some embodiments, the coordinate determination model includes a part segmentation layer, a contour coordinate determination layer, a rectangular box coordinate determination layer, and a two-dimensional coordinate range determination layer. The input of the part segmentation layer is the image of the facial mask under white light illumination. The outputs of the part segmentation layer are the overall image of the facial mask, the image of the left eye part of the facial mask, the image of the right eye part of the facial mask, the image of the nose part of the facial mask, and the image of the mouth part of the facial mask. The input of the contour coordinate determination layer is the overall image of the facial mask, the image of the left eye part of the facial mask, the image of the right eye part of the facial mask, the image of the nose part of the facial mask, and the image of the mouth part of the facial mask. The outputs of the contour coordinate determination layer are the coordinates of multiple points on the overall contour of the facial mask, the coordinates of multiple points on the contour of the left eye part of the facial mask, the coordinates of multiple points on the contour of the right eye part of the facial mask, the coordinates of multiple points on the contour of the nose part of the facial mask, and the coordinates of multiple points on the contour of the mouth part of the facial mask. The input of the rectangular box coordinate determination layer is the coordinates of multiple points on the overall contour of the facial mask, the coordinates of multiple points on the contour of the left eye part of the facial mask, the coordinates of multiple points on the contour of the right eye part of the facial mask, the coordinates of multiple points on the contour of the nose part of the facial mask, and the coordinates of multiple points on the contour of the mouth part of the facial mask. The outputs of the rectangular box coordinate determination layer are the coordinate ranges of multiple small rectangular boxes within the overall contour of the facial mask, the coordinate ranges of multiple small rectangular boxes within the contour of the left eye part of the facial mask, the coordinate ranges of multiple small rectangular boxes within the contour of the right eye part of the facial mask, the coordinate ranges of multiple small rectangular boxes within the contour of the nose part of the facial mask, and the coordinate ranges of multiple small rectangular boxes within the contour of the mouth part of the facial mask. The input of the two-dimensional coordinate range determination layer is the coordinate ranges of multiple small rectangular boxes within the overall contour of the facial mask, the coordinate ranges of multiple small rectangular boxes within the contour of the left eye part of the facial mask, the coordinate ranges of multiple small rectangular boxes within the contour of the right eye part of the facial mask, the coordinate ranges of multiple small rectangular boxes within the contour of the nose part of the facial mask, and the coordinate ranges of multiple small rectangular boxes within the contour of the mouth part of the facial mask. The outputs of the two-dimensional coordinate range determination layer are the two-dimensional coordinate ranges of the overall facial mask, the two-dimensional coordinate ranges of the left eye part of the facial mask, the two-dimensional coordinate ranges of the right eye part of the facial mask, the two-dimensional coordinate ranges of the nose part of the facial mask, and the two-dimensional coordinate ranges of the mouth part of the facial mask.

[0030] The coordinates of multiple points on the overall contour of the facial mask refer to the position coordinates of multiple points on the overall contour of the facial mask. These coordinates can be calculated by the contour coordinate determination layer, and the coordinates of multiple points on the overall contour of the facial mask can be used to describe the overall contour of the facial mask.

[0031] The coordinates of multiple points on the contour of the left eye part of the facial mask, the coordinates of multiple points on the contour of the right eye part of the facial mask, the coordinates of multiple points on the contour of the nose part of the facial mask, and the coordinates of multiple points on the contour of the mouth part of the facial mask. These coordinates are used to describe the position coordinates of the points corresponding to the contour shapes of the respective parts.

[0032] The coordinate ranges of multiple small rectangular frames within the overall contour of the facial mask refer to the coordinate ranges of multiple small rectangular frames divided within the overall contour of the facial mask. The multiple small rectangular frames are used to describe the position information of different regions inside the facial mask. The coordinate ranges of multiple small rectangular frames within the contour of the left-eye part of the facial mask refer to the coordinate ranges of multiple small rectangular frames divided within the contour of the left-eye part of the facial mask. The multiple small rectangular frames are used to describe the position information of the left-eye part of the facial mask. The coordinate ranges of multiple small rectangular frames within the contour of the right-eye part of the facial mask refer to the coordinate ranges of multiple small rectangular frames divided within the contour of the right-eye part of the facial mask. The multiple small rectangular frames are used to describe the position information of the right-eye part of the facial mask. The coordinate ranges of multiple small rectangular frames within the contour of the nose part of the facial mask refer to the coordinate ranges of multiple small rectangular frames divided within the contour of the nose part of the facial mask. The multiple small rectangular frames are used to describe the position information of the nose part of the facial mask. The coordinate ranges of multiple small rectangular frames within the contour of the mouth part of the facial mask refer to the coordinate ranges of multiple small rectangular frames divided within the contour of the mouth part of the facial mask. The multiple small rectangular frames are used to describe the position information of the mouth part of the facial mask.

[0033] As an example, assuming that the nose area consists of multiple irregular curves or boundaries, we can divide it into multiple small continuous rectangular regions, and then use the two-dimensional coordinate ranges of each small rectangular region to describe the two-dimensional coordinate range of the nose part of the facial mask. For each small rectangular region, its two-dimensional coordinate range can be represented in the form of a rectangular frame. Suppose the left boundary of a certain small region is x1, the right boundary is x2, the upper boundary is y1, and the lower boundary is y2. Then the coordinate range of this small region can be expressed as [x1, x2] × [y1, y2]. By combining the two-dimensional coordinate ranges of all small regions, the description of the entire nose area can be obtained. The coordinate ranges of each small region can be put into a list, which represents the coordinate range occupied by the nose area. The two-dimensional coordinate range determination layer obtains the description of the entire nose area by combining the two-dimensional coordinate ranges of all small rectangular regions. The two-dimensional coordinate range determination layer can put the coordinate ranges of each small region into a list, and this list represents the coordinate range occupied by the eye area.

[0034] The part segmentation layer is used to segment the mask image into five parts, namely the whole, left eye, right eye, nose, and mouth, for subsequent processing. The contour coordinate determination layer is used to determine the point coordinates of the contours of each part, the rectangular box coordinate determination layer is used to determine the coordinate ranges of multiple small rectangular boxes within each part, and the two-dimensional coordinate range determination layer is used to merge the coordinate ranges of all small rectangular boxes to obtain the two-dimensional coordinate range of the entire part. Dividing the coordinate determination model into multiple layers allows each layer to focus on different tasks and improves the accuracy of the model. For example, the part segmentation layer is specifically used to segment the entire image into each part, while the contour coordinate determination layer focuses on determining the contours of each part. In addition, layering also facilitates the interpretability and debugging of the model.

[0035] Step S3, construct multiple nodes and multiple edges between the multiple nodes. The multiple nodes include a left eye node, a right eye node, a nose node, and a mouth node. The node feature of the left eye node is the two-dimensional coordinate range of the left eye part of the mask, the node feature of the right eye node is the two-dimensional coordinate range of the right eye part of the mask, the node feature of the nose node is the two-dimensional coordinate range of the nose part of the mask and the two-dimensional coordinate range of the whole mask, and the node feature of the mouth node is the two-dimensional coordinate range of the mouth part of the mask.

[0036] Graph-structured data is a data structure composed of nodes and edges. The graph structure includes multiple nodes and multiple edges between the multiple nodes. The graph neural network model is a neural network that directly acts on graph-structured data.

[0037] By representing different parts of the mask as nodes, the structure of the mask can be organized and understood more clearly. Each node represents a specific part, making the analysis of the mask shape more interpretable.

[0038] In some embodiments, the multiple edges between the multiple nodes represent the positional relationship between the nodes, and the edge features of the multiple edges represent the shortest distance between the nodes and the direction corresponding to the shortest distance. By connecting nodes with edges, the positional relationship between each part can be established. Such modeling can help with further facial analysis and feature extraction.

[0039] Construct corresponding nodes according to the characteristics of each part and connect them with edges. This can be achieved through the graph data structure in graph theory, where each node represents a part and the edge represents the relationship between different parts. The node features can be stored as node attributes in the graph, and the edges can be used to represent the positional relationship between the nodes.

[0040] Step S4, process the multiple nodes and the multiple edges between the multiple nodes based on the graph neural network model to obtain whether the mask shape is compliant.

[0041] The graph neural network model includes a graph neural network (GNN) and a fully connected layer. The graph neural network model is used to process multiple nodes and multiple edges between the multiple nodes of the facial mask to determine whether the shape of the facial mask complies with the regulations.

[0042] In the analysis of the facial mask shape, the nodes represent different parts of the facial mask, such as the left eye, the right eye, the nose, and the mouth, etc. Each node has specific node features, such as a two-dimensional coordinate range. Through the nodes, each part of the facial mask can be modeled and analyzed.

[0043] The graph neural network model can learn the global features of the facial mask, including information such as the relative positions, shapes, and sizes of each part. Through global feature learning, it can be more comprehensively determined whether the shape of the facial mask complies with the regulations. By processing multiple nodes and multiple edges, the correlation relationships between each part of the facial mask can be captured. For example, the position of the nose should be between the two eyes, and the position of the mouth should be below the nose, etc. By modeling and processing these nodes and edges, it can be more accurately determined whether the shape of the facial mask and the morphological distribution of each part comply with the regulations. In some embodiments, a labeled dataset of compliant and non-compliant facial masks can be used to train the graph neural network model. Then, the graph neural network model can make inferences on new facial mask data and determine whether the shape of the facial mask complies with the regulations.

[0044] The input of the graph neural network model is the multiple nodes and the multiple edges between the multiple nodes, and the output of the graph neural network model is that the facial mask shape is compliant or the facial mask shape is non-compliant. If the output of the graph neural network model is that the facial mask shape is non-compliant, the non-compliance information can be sent to the management terminal.

[0045] In some embodiments, the quality of the facial mask can also be detected by a light treatment model to determine whether there are cracks in the facial mask. Figure 2 FIG. is a schematic flowchart of detecting the quality of a facial mask by a light treatment model provided by an embodiment of the present invention. The detecting the quality of a facial mask by a light treatment model includes steps S21 to S24: Step S21, obtaining multiple images of the facial mask under multiple colored lights.

[0046] The multiple images of the facial mask under multiple colored lights include multiple images of the facial mask under red light illumination, multiple images of the facial mask under green light illumination, and multiple images of the facial mask under blue light illumination.

[0047] Multiple images of the facial mask under multiple colored lights can be obtained by taking pictures with a camera or other image acquisition devices under the set red light, green light, and blue light illuminations.

[0048] Step S22: Use the light processing model to determine multiple masks to be verified based on multiple images of each mask under multiple colored lights.

[0049] The multiple masks to be verified refer to multiple masks that need to be further tested by the stretching test.

[0050] Red light, green light, and blue light have different wavelengths. When light of different wavelengths shines on the mask, the mask material will have different reflections and absorptions of the light. These reflection and absorption phenomena will present different image features, thus affecting the detection of cracks. Specifically, when light of different wavelengths shines on the mask, the color and contrast on the mask surface will change. The wavelength of red light is relatively long and can penetrate deeper into the layers of the mask, so it can better detect cracks in the deeper layers of the mask. The wavelength of green light is moderate and can reflect the crack conditions on the surface and middle layers of the mask. The wavelength of blue light is relatively short and can reflect the tiny cracks on the mask surface. By using lights of different wavelengths to shine on the mask, the crack conditions on the surface and inside of the mask can be captured more comprehensively, improving the accuracy and effectiveness of crack detection.

[0051] The convolutional neural network can capture the spatial features in multiple images of each mask under multiple colored lights through convolutional layers and pooling layers, so as to detect possible cracks. The convolutional neural network can be trained to learn the patterns and differences between the normal area and the abnormal area of the mask. By training on a large number of mask images with known cracks, the convolutional neural network can learn the features of these cracks and determine multiple masks to be verified.

[0052] The light processing model can analyze and compare images under different lighting conditions to determine multiple masks to be verified that need further verification.

[0053] The light processing model is a convolutional neural network model. The input of the light processing model is multiple images of the mask under multiple colored lights, and the output of the light processing model is whether there are cracks.

[0054] Step S23: Obtain the stretching video of each mask to be verified.

[0055] In some embodiments, a special stretching test instrument, such as a tensile testing machine, can be used to stretch each mask to be verified, and at the same time, a video of the mask to be verified during the stretching process is taken to obtain the stretching video of each mask to be verified.

[0056] Step S24: Use the stretching video processing model to determine whether there are cracks in each mask to be verified based on the stretching video of each mask to be verified.

[0057] The stretching video processing model is a gated recurrent unit. The input of the stretching video processing model is the stretching video of each mask to be verified, and the output of the stretching video processing model is whether there are cracks in each mask to be verified or there are no cracks in each mask to be verified.

[0058] The gated recurrent unit (GRU) is used to process sequence data and temporal information. The gated recurrent unit consists of three components: a memory unit, an update gate, and a reset gate. The memory unit stores information from previous time steps and passes it to the input and hidden state of the current time step to retain historical information. The update gate determines whether to update the information in the memory unit. It controls the degree of update of the memory unit based on the input of the current time step and the hidden state of the previous time step. When the update gate approaches 1, more past information will be retained; when the update gate approaches 0, more new information will be passed. The reset gate determines how to utilize the memory unit of the previous time step. It decides which information to discard based on the input of the current time step and the hidden state of the previous time step. The role of the reset gate is to better adapt to the current input. Through the adjustment of the update gate and the reset gate, the gated recurrent unit effectively solves the problem of gradient disappearance, enabling the model to better handle the dependency relationships of long sequence data.

[0059] The gated recurrent unit can process the stretching videos of each mask to be verified in consecutive time periods and comprehensively consider the characteristics of the correlation relationships between the stretching videos of each mask to be verified at each time point, making the characteristics of the output more accurate and comprehensive.

[0060] Multiple images under various colored lights can provide observations of different angles and details of the mask surface, but this method also has certain limitations. For example, using a specific wavelength light source may mask some fine cracks, resulting in missed detections. During the stretching process, the original cracks on the mask will deform. These deformations can be captured through consecutive frames, so the stretching video can more accurately obtain the crack information of the mask. The mask may undergo minor deformations and cracks during the stretching process, and these changes are difficult to capture in static images. By analyzing the dynamic changes of the mask between consecutive frames through the gated recurrent unit, the dynamic characteristics of the mask can be learned, thereby enhancing the ability to detect cracks.

[0061] In some embodiments, the stretching video processing model includes a crack location layer, a crack stretching video segment segmentation layer, and a crack determination layer. The input of the crack location layer is the stretching video of each mask to be verified, and the output of the crack location layer is the suspected crack location of each mask to be verified. The input of the crack stretching video segment segmentation layer is the suspected crack location of each mask to be verified and the stretching video of each mask to be verified, and the output of the crack stretching video segment segmentation layer is the stretched and segmented video of the suspected crack location of each mask to be verified. The input of the crack determination layer is the stretched and segmented video of the suspected crack location of each mask to be verified, and the output of the crack determination layer is that there is a crack or there is no crack in each mask to be verified. By dividing the stretching video processing model into multiple layers, the information of the crack can be gradually extracted, analyzed, and determined, so as to realize the crack detection of the mask to be verified. Each layer undertakes a specific task and cooperates with each other to complete the whole processing process. Such a hierarchical design can improve the robustness and efficiency of the system.

[0062] Based on the same inventive concept, Figure 3 FIG. is a schematic diagram of a big data-based cosmetic quality detection system provided by an embodiment of the present invention. The big data-based cosmetic quality detection system includes: An acquisition module 31, configured to acquire an image of the mask under white light illumination; A coordinate determination module 32, configured to use a coordinate determination model based on the image of the mask under white light illumination to determine the two-dimensional coordinate range of the whole mask, the two-dimensional coordinate range of the left eye part of the mask, the two-dimensional coordinate range of the right eye part of the mask, the two-dimensional coordinate range of the nose part of the mask, and the two-dimensional coordinate range of the mouth part of the mask; A construction module 33, configured to construct multiple nodes and multiple edges between the multiple nodes. The multiple nodes include a left eye node, a right eye node, a nose node, and a mouth node. The node feature of the left eye node is the two-dimensional coordinate range of the left eye part of the mask, the node feature of the right eye node is the two-dimensional coordinate range of the right eye part of the mask, the node feature of the nose node is the two-dimensional coordinate range of the nose part of the mask and the two-dimensional coordinate range of the whole mask, and the node feature of the mouth node is the two-dimensional coordinate range of the mouth part of the mask; A compliance judgment module 34, configured to process the multiple nodes and the multiple edges between the multiple nodes based on a graph neural network model to obtain whether the shape of the mask is compliant.

Claims

1. A cosmetics quality detection method based on big data, characterized in that: include: Acquire an image of the mask under white light; Based on the image of the facial mask under white light, a coordinate determination model is used to determine the two-dimensional coordinate range of the entire facial mask, the two-dimensional coordinate range of the left eye portion of the facial mask, the two-dimensional coordinate range of the right eye portion of the facial mask, the two-dimensional coordinate range of the nose portion of the facial mask, and the two-dimensional coordinate range of the mouth portion of the facial mask; Constructing a plurality of nodes and a plurality of edges between the plurality of nodes, the plurality of nodes comprising a left eye node, a right eye node, a nose node, and a mouth node, the node feature of the left eye node being the two-dimensional coordinate range of the left eye part of the mask, the node feature of the right eye node being the two-dimensional coordinate range of the right eye part of the mask, the node feature of the nose node being the two-dimensional coordinate range of the nose part of the mask and the two-dimensional coordinate range of the entire mask, and the node feature of the mouth node being the two-dimensional coordinate range of the mouth part of the mask; Based on the graph neural network model, the multiple nodes and the multiple edges between the multiple nodes are processed to obtain whether the shape of the mask is compliant.

2. The cosmetics quality detection method based on big data according to claim 1, characterized in that: The method further comprises: Acquire multiple images of multiple masks under multiple colored lights; Determine multiple masks to be verified using a light processing model based on multiple images of each mask under multiple colors of light; Get the stretching video of each mask to be verified; Based on the stretching video of each of the facial masks to be verified, a stretching video processing model is used to determine whether each of the facial masks to be verified has cracks.

3. The cosmetics quality detection method based on big data according to claim 2, characterized in that: The light processing model is a convolutional neural network model, the input of the light processing model is a plurality of images of the facial mask under illumination of a variety of colored lights, and the output of the light processing model is a plurality of facial masks to be verified.

4. The cosmetics quality detection method based on big data according to claim 3, characterized in that: The multiple images of the facial mask under multiple colored lights include multiple images of the facial mask under red light, multiple images of the facial mask under green light, and multiple images of the facial mask under blue light.

5. The cosmetics quality detection method based on big data according to claim 1, characterized in that: The coordinate determination model is a convolutional neural network model, the input of the coordinate determination model is an image of the facial mask under white light, and the output of the coordinate determination model is the two-dimensional coordinate range of the entire facial mask, the two-dimensional coordinate range of the left eye of the facial mask, the two-dimensional coordinate range of the right eye of the facial mask, the two-dimensional coordinate range of the nose of the facial mask, and the two-dimensional coordinate range of the mouth of the facial mask.

6. A cosmetics quality detection system based on big data, characterized in that: include: An acquisition module, used for acquiring an image of the mask under white light; A coordinate determination module, for determining the two-dimensional coordinate range of the entire mask, the two-dimensional coordinate range of the left eye of the mask, the two-dimensional coordinate range of the right eye of the mask, the two-dimensional coordinate range of the nose of the mask, and the two-dimensional coordinate range of the mouth of the mask using a coordinate determination model based on the image of the mask under white light; A construction module, for constructing a plurality of nodes and a plurality of edges between the plurality of nodes, the plurality of nodes comprising a left eye node, a right eye node, a nose node, and a mouth node, the node feature of the left eye node being the two-dimensional coordinate range of the left eye portion of the mask, the node feature of the right eye node being the two-dimensional coordinate range of the right eye portion of the mask, the node feature of the nose node being the two-dimensional coordinate range of the nose portion of the mask and the two-dimensional coordinate range of the entire mask, and the node feature of the mouth node being the two-dimensional coordinate range of the mouth portion of the mask; The compliance judgment module is used to process the multiple nodes and the multiple edges between the multiple nodes based on the graph neural network model to determine whether the shape of the mask is compliant.

7. The cosmetics quality detection system based on big data according to claim 6, characterized in that: The system is also used to: Acquire multiple images of multiple masks under multiple colored lights; Determine multiple masks to be verified using a light processing model based on multiple images of each mask under multiple colors of light; Get the stretching video of each mask to be verified; Based on the stretching video of each of the facial masks to be verified, a stretching video processing model is used to determine whether each of the facial masks to be verified has cracks.

8. The cosmetics quality detection system based on big data according to claim 7, characterized in that: The light processing model is a convolutional neural network model, the input of the light processing model is a plurality of images of the facial mask under illumination of a variety of colored lights, and the output of the light processing model is a plurality of facial masks to be verified.

9. The cosmetics quality detection system based on big data according to claim 8, characterized in that: The multiple images of the facial mask under multiple colored lights include multiple images of the facial mask under red light, multiple images of the facial mask under green light, and multiple images of the facial mask under blue light.

10. The cosmetics quality detection system based on big data according to claim 6, characterized in that: The coordinate determination model is a convolutional neural network model, the input of the coordinate determination model is an image of the facial mask under white light, and the output of the coordinate determination model is the two-dimensional coordinate range of the entire facial mask, the two-dimensional coordinate range of the left eye of the facial mask, the two-dimensional coordinate range of the right eye of the facial mask, the two-dimensional coordinate range of the nose of the facial mask, and the two-dimensional coordinate range of the mouth of the facial mask.

Images