Intelligent tea tasting system based on intelligent recognition of computer vision

Abstract

The invention discloses an intelligent tea tasting system based on intelligent recognition of computer vision. The system comprises a moving module used for moving a tea photographing end; the photographing module is used for photographing tea leaves needing to be recognized; the uploading module is used for uploading tea shooting data; the storage module is used for storing and calling a tea database; the query module is used for photographing and uploading pictures for query, and the recognition module is used for judging the type of tea leaves; the invention relates to the technical field of intelligent tea tasting and identification. According to the intelligent tea tasting system for intelligent identification based on computer vision, the color correction module is arranged, the color correction module adopts color correction algorithms such as white balance, histogram equalization and adaptive histogram equalization in a preprocessing stage, and color distribution of an input image is automatically optimized so as to reduce ambient light interference; the color cast problem generated by the mobile phone photographed image is corrected, and the extraction module is used for extracting key features of the collected image.

Description

Technical Field

[0001] The present invention relates to the technical field of intelligent tea tasting, and in particular to an intelligent tea tasting system based on computer vision and intelligent recognition. Background Art

[0002] Traditional tea classification mainly relies on manual sensory evaluation, but this method is slow and easily affected by the subjective factors of tea tasters, making it difficult to meet the needs of large-scale rapid tea classification. For example, during the peak tea harvest season, a large amount of tea needs to be classified and put on the market in a timely manner, and the efficiency of manual review cannot keep up. Referring to the authorization announcement number "CN118410220A", an intelligent tea tasting and recommendation system and equipment is provided. The system includes: a data acquisition module for collecting first information related to the user, and the first information is configured to be related to tea tasting; a matching module for processing and analyzing the collected first information;

[0003] As shown in the above patent, the existing intelligent tea tasting system based on computer vision intelligent recognition is used by mixing traditional CV and simple CNN architecture algorithms, mainly relying on manual features or shallow networks, with general generalization and easy to be affected by the environment. Traditional image processing requires standard light sources, has strong limitations and weak scalability. For the tea market, the market supervision ability is limited and the sampling coverage rate is low. For consumers, due to the lack of professional tools, manual subjectivity is strong, it is easily affected by the environment, and the recognition accuracy fluctuates greatly. Summary of the Invention

[0004] In response to the shortcomings of the existing technology, the present invention provides an intelligent tea tasting system based on computer vision and intelligent recognition, which solves the problems that the system mainly relies on manual features or shallow networks, has poor generalization, and is easily affected by the environment.

[0005] To achieve the above objectives, the present invention is implemented through the following technical solutions: a tea intelligent tasting system based on computer vision intelligent recognition, including a mobile module for moving the tea photography end;

[0006] A photo module is used to take pictures of the tea leaves to be identified;

[0007] Upload module, used to upload the data of photographed tea leaves;

[0008] Storage module, used for tea database storage and call processing;

[0009] Query module, used to query pictures uploaded by taking photos;

[0010] Identification module, used to determine the type of tea;

[0011] Processing module, used for classification and grading of tea types;

[0012] A feedback module, used to feed back the photographed tea leaves processing data to the mobile module;

[0013] Add a module to transfer the data that cannot be recognized by the recognition module to the storage module for storage.

[0014] Preferably, the photographing module and the uploading module are respectively arranged in the mobile module, and the query module is respectively connected to the uploading module and the storage module.

[0015] Preferably, the identification module is connected to the query module, the adding module and the processing module respectively, and the feedback module is connected to the moving module and the processing module respectively.

[0016] Preferably, the storage module is connected to the adding module and the processing module respectively, the photographing module includes a color correction module, an image acquisition module, and an extraction module, the extraction module is connected to the uploading module, and the color correction module, image acquisition module, and extraction module are all arranged in the mobile module.

[0017] Preferably, the color correction module is used to reduce ambient light interference, the extraction module is used to extract key features of the captured image, and the image acquisition module is connected to the color correction module and the extraction module respectively.

[0018] Preferably, the processing module includes a classification module, a grading module, and an imaging module. The grading module is connected to the classification module and the imaging module respectively. The classification module is connected to the recognition module. The grading module and the classification module are both connected to the feedback module and the storage module.

[0019] Preferably, the adding module includes an acquisition module, a taste module, a physical detection module, and an analysis module. The acquisition module is used to collect tea leaves photographed by the photographing module. The taste module, the olfactory module, and the physical detection module are respectively connected to the acquisition module and the analysis module. The analysis module is connected to the storage module.

[0020] Preferably, the taste module includes an electronic tongue, the olfactory module includes an electronic nose, and the physical detection module includes a colorimeter and a texture analyzer.

[0021] Beneficial effects

[0022] The present invention provides a tea tasting system based on computer vision and intelligent recognition. Compared with the existing technology, it has the following advantages:

[0023] (1) The intelligent tea tasting system based on computer vision and intelligent recognition is implemented by setting a processing module and a photographing module in the system. The processing module adopts the ResNet50 model and adds CBAM to the ResNet50 basic model so that the model focuses on the key features in the image data. By combining CBAM and the final confusion matrix, accurate and efficient identification of tea types and grades can be achieved. The color correction module adopts color correction algorithms such as white balance, histogram equalization, and adaptive histogram equalization in the preprocessing stage to automatically optimize the color distribution of the input image to reduce ambient light interference and correct the color cast problem caused by mobile phone camera images. The extraction module is used to extract the key features of the captured image. The image acquisition module is connected to the color correction module and the extraction module respectively. This setting deploys a lightweight model, supports real-time image acquisition and preliminary feature extraction, reduces cloud computing pressure, and is connected in series with the cloud server through the API interface to realize data recognition and distributed computing processing, reduce database query delay, and improve response speed.

[0024] (2) The intelligent tea tasting system based on computer vision and intelligent recognition is implemented by setting a grading module in the system. The grading module uses CNN (convolutional neural network) for tea classification. The classification module uses multispectral or hyperspectral imaging to combine spectral data of different bands to improve recognition accuracy. The shaping module uses laser scanning or structured light imaging to obtain 3D shape data of tea leaves and analyze indicators such as the tightness and uniformity of the tea leaves. Microscopic image analysis is used to observe the cell structure of tea leaves to assist in identifying authenticity.

[0025] (3) The intelligent tea tasting system based on computer vision and intelligent recognition sets up a collection module, a taste module, a physical detection module, and an analysis module in the system. The collection module collects the tea that cannot be identified and allows the electronic tongue and electronic nose in the taste module to simulate human taste and smell respectively to detect the taste and aroma components of the tea (such as amino acids, tea polyphenols, and volatile substances). The colorimeter and texture analyzer are used to quantify the physical properties of the tea, such as color, shape, and hardness, to reduce subjective judgment bias and prevent inconsistent judgments on the same tea due to the experience and preferences of different tea tasters. BRIEF DESCRIPTION OF THE DRAWINGS

[0026] Figure 1 This is a main principle block diagram of the present invention;

[0027] Figure 2 It is the main flow chart of the present invention;

[0028] Figure 3 This is a principle block diagram of the loss analysis module of the present invention;

[0029] Figure 4 This is a principle block diagram of the monitoring module in the present invention;

[0030] Figure 5 This is a principle block diagram of the blood transfusion module of the present invention. DETAILED DESCRIPTION

[0031] The following will clearly and completely describe the technical solutions in the embodiments of the present invention in conjunction with the accompanying drawings. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without making creative efforts are within the scope of protection of the present invention.

[0032] like Figure 1-3 The first embodiment shown is a tea intelligent tasting system based on computer vision intelligent recognition, including a mobile module for moving the tea photography terminal; a photography module for photographing the tea to be identified; an upload module for uploading the photographed tea data; a storage module for storing and processing tea database calls and a built-in API interface; a query module for photographing and uploading images for query; and an identification module for determining the type of tea.

[0033] The processing module is used for tea type classification and grading. The processing module uses the ResNet50 model and adds CBAM to the ResNet50 base model to focus on key features in the image data, better for fine-grained image classification, and realize the identification of tea categories and grades. At the same time, custom branches are added to facilitate better training data and strengthen features. The ResNet50 base model takes a 224×224×3 image as input, and after 7×7 convolution (64 kernels) and 3×3 maximum pooling, it enters four residual stages:

[0034] Stage 1: 3 residual blocks (1×1 dimensionality reduction → 3×3 convolution → 1×1 dimensionality increase + residual connection), output 56×56×256;

[0035] Stage 2: 4 residual blocks, output 28×28×512;

[0036] Stage 3: 6 residual blocks, output 14×14×1024;

[0037] Stage 4: 3 residual blocks, output 7×7×2048.

[0038] Finally, it is connected to the custom branch through the global pooling layer. The addition of CBAM and custom branches effectively reduces the requirement for the amount of training data, greatly shortens the training time, and improves the training accuracy. In traditional CNN, information propagates from the input through each layer, which will cause the gradient to vanish, making it difficult to train deep neural networks. The ResNet network solves this problem by adding residual blocks. Each residual block contains multiple convolutional layers and a "shortcut" across several layers, allowing information to be directly transmitted from a previous layer to a later layer. This design avoids the gradient vanishing problem during deep network training and makes the network easier to train.

[0039] The feedback module is used to feed back the processed data of the photographed tea leaves to the mobile module; the adding module is used to transfer the data that the recognition module cannot recognize to the storage module for storage; the photographing module and the uploading module are respectively arranged in the mobile module, and the query module is respectively connected to the uploading module and the storage module; the recognition module is respectively connected to the query module, the adding module, and the processing module, and the feedback module is respectively connected to the mobile module and the processing module;

[0040] The storage module is connected to the adding module and the processing module respectively. The photographing module includes a color correction module, an image acquisition module, and an extraction module. The extraction module is connected to the uploading module. The color correction module, the image acquisition module, and the extraction module are all set in the mobile module. The storage module has a self-built tea image library. The data was collected from January 14 to 15, 2025, and the acquisition method was to take pictures with a mobile phone under natural light. Data characteristics: Three typical tea samples of black tea, green tea and yellow tea were selected from the six major tea categories. One tea product was selected from each tea sample for data collection; Data scale: The three tea products are Keemun black tea, Maofeng tea, and yellow tea. Keemun black tea has seven grades, with ninety pictures for each grade; Maofeng has seven grades, with ninety pictures for each grade; Yellow tea has nine grades, with seventy pictures for each grade. The total sample size of the data set reaches 1,800;

[0041] The color correction module uses color correction algorithms such as white balance, histogram equalization, and adaptive histogram equalization in the preprocessing stage to automatically optimize the color distribution of the input image to reduce ambient light interference. The extraction module is used to extract key features of the acquired image. The image acquisition module is connected to the color correction module and the extraction module respectively. This setup deploys a lightweight model, supports real-time image acquisition and preliminary feature extraction, reduces cloud computing pressure, and connects in series with the cloud server through the API interface to achieve data recognition and distributed computing processing, reduce database query latency, and improve response speed;

[0042] like Figure 1-4 The second embodiment shown is mainly different from the first embodiment in that:

[0043] The processing module includes a classification module, a grading module, and an imaging module. The grading module is connected to the classification module and the imaging module respectively, and the classification module is connected to the recognition module. The grading module and the classification module are both connected to the feedback module and the storage module. The grading module uses CNN (convolutional neural network) for tea classification. The classification module uses multispectral or hyperspectral imaging to combine spectral data from different bands to improve recognition accuracy. The shaping module uses laser scanning or structured light imaging to obtain 3D shape data of tea leaves and analyze indicators such as tea strand tightness and evenness. Microscopic image analysis: observes the cell structure of tea leaves to assist in identifying authenticity.

[0044] like Figure 1-5 The third embodiment shown is mainly different from the second embodiment in that:

[0045] The added module includes a collection module, a taste module, a physical detection module, and an analysis module. The collection module is used to collect tea leaves photographed by the camera module. The taste module, the olfactory module, and the physical detection module are connected to the collection module and the analysis module respectively. The analysis module is connected to the storage module.

[0046] The taste module includes an electronic tongue, the smell module includes an electronic nose, and the physical detection module includes a colorimeter and a texture analyzer. The electronic tongue and electronic nose simulate human taste and smell respectively to detect the flavor and aroma components of tea (such as amino acids, tea polyphenols, and volatile substances). The colorimeter and texture analyzer are used to quantify the physical properties of tea, such as color, shape, and hardness, to reduce subjective judgment bias.

[0047] During use, the user takes a photo of the tea to be tasted through the mobile system to identify the tea. The photographing system uses the color correction module to reduce the impact of ambient light on the photographed image. The image acquisition module collects and processes the color-corrected image. The extraction module extracts the key features in the image and uploads it to the query module through the upload module. The query module stores the tea encyclopedia in the module database and identifies the tea through the recognition module. When the tea can be classified, the classification module inside the processing module classifies the tea varieties. After the tea varieties are classified, the grading module classifies the tea grades through target detection and spectral imaging. The imaging module obtains the 3D shape data of the tea through laser scanning or structured light imaging, analyzes indicators such as the tightness and uniformity of the tea leaves, and observes the cell structure of the tea through microscopic image analysis. , assist in identifying authenticity, and transmit the detection data to the mobile terminal through the feedback module to realize intelligent tasting of tea. When the tea cannot be classified, the system will collect the tea that is photographed and cannot be identified through the collection module, and detect the taste and color, shape, hardness and other physical properties of the tea through the olfactory module, taste module, and physical detection module respectively. The analysis module analyzes and summarizes the above data, renames the tea of ​​this data and adds it to the database of the storage module to establish a tea intelligent identification information library. Through data processing training test and image recognition test, the training test mainly verifies the training accuracy of the model for different types and grades of tea. The recognition test focuses on the recognition accuracy, stability and response speed of the system under high concurrency, as well as the calculation and processing of data.

[0048] At the same time, the contents not described in detail in this specification belong to the existing technology known to those skilled in the art, and the model parameters of each electrical appliance are not specifically limited, and conventional equipment can be used.

[0049] It should be noted that, in this document, relational terms such as first and second, etc., are used only to distinguish one entity or operation from another entity or operation, and do not necessarily require or imply any actual relationship or order between these entities or operations. Moreover, the terms "comprises," "comprising," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that includes a list of elements includes not only those elements but also other elements not explicitly listed, or elements inherent to such process, method, article, or apparatus.

[0050] While embodiments of the present invention have been shown and described, it will be appreciated by those skilled in the art that various changes, modifications, substitutions, and variations may be made to these embodiments without departing from the principles and spirit of the invention, and that the scope of the invention is defined by the appended claims and their equivalents.

Claims

1. A tea tasting system based on computer vision and intelligent recognition, including a mobile module for moving a tea photography terminal, is characterized by: A photo module is used to take pictures of the tea leaves to be identified; Upload module, used to upload the data of photographed tea leaves; Storage module, used for tea database storage and call processing; Query module, used to query pictures uploaded by taking photos; Identification module, used to determine the type of tea; Processing module, used for classification and grading of tea types; A feedback module, used to feed back the photographed tea leaves processing data to the mobile module; Add a module to transfer the data that cannot be recognized by the recognition module to the storage module for storage.

2. The intelligent tea tasting system based on computer vision and intelligent recognition according to claim 1, characterized in that: The photographing module and the uploading module are respectively arranged in the moving module, and the query module is respectively connected with the uploading module and the storage module.

3. The intelligent tea tasting system based on computer vision and intelligent recognition according to claim 1, characterized in that: The identification module is connected to the query module, the adding module and the processing module respectively, and the feedback module is connected to the moving block and the processing module respectively.

4. The intelligent tea tasting system based on computer vision and intelligent recognition according to claim 1, characterized in that: The storage module is connected to the adding module and the processing module respectively. The photographing module includes a color correction module, an image acquisition module, and an extraction module. The extraction module is connected to the uploading module. The color correction module, image acquisition module, and extraction module are all arranged in the mobile module.

5. The intelligent tea tasting system based on computer vision and intelligent recognition according to claim 4, characterized in that: The color correction module is used to reduce ambient light interference, and the extraction module is used to extract key features of the captured image. The image acquisition module is connected to the color correction module and the extraction module respectively.

6. The intelligent tea tasting system based on computer vision and intelligent recognition according to claim 1, characterized in that: The processing module includes a classification module, a grading module, and an imaging module. The grading module is connected to the classification module and the imaging module respectively. The classification module is connected to the recognition module. The grading module and the classification module are both connected to the feedback module and the storage module.

7. The intelligent tea tasting system based on computer vision and intelligent recognition according to claim 1, characterized in that: The adding module includes a collection module, a taste module, a physical detection module, and an analysis module. The collection module is used to collect tea leaves photographed by the photography module. The taste module, the olfactory module, and the physical detection module are respectively connected to the collection module and the analysis module. The analysis module is connected to the storage module.

8. The intelligent tea tasting system based on computer vision and intelligent recognition according to claim 7, characterized in that: The taste module includes an electronic tongue, the olfactory module includes an electronic nose, and the physical detection module includes a colorimeter and a texture analyzer.

Images