Entry-exit directory management system for identifying types of infectious substances through images

Abstract

The invention relates to the technical field of entry and exit directory management systems, and provides an entry and exit directory management system for identifying infectious substance categories through images, in the aspect of image identification, an algorithm fuses CNN and an attention mechanism and combines manual features, image features can be extracted more accurately, and the accuracy of the image identification is improved. Therefore, accurate classification and danger level judgment are carried out on infectious substances, misjudgment conditions are reduced, the accuracy and reliability of recognition are effectively improved, the response measure execution module is linked with the early warning notification module, timely execution of prevention and control measures is ensured, efficient early warning is carried out to related departments, the emergency processing capacity and collaboration are enhanced, and the emergency treatment efficiency is improved. The front end and the rear end of the whole framework and the process design are tightly matched, the work process forms a closed loop, the operation convenience of port workers is improved, powerful data support can be provided for management decisions, the safety and efficient management of ports are powerfully guaranteed, and the system has outstanding advantages in the field of infectious substance management and control.

Description

Technical Field

[0001] The present invention relates to the technical field of entry-exit list management systems, and in particular to an entry-exit list management system for identifying categories of infectious substances through images. Background Art

[0002] Against the backdrop of accelerating globalization, the flow of people and goods entering and leaving the country has increased dramatically, and the risk of cross-border transmission of infectious substances has become increasingly prominent. Traditional methods of infectious substance detection and management mainly rely on manual inspection and limited laboratory testing, which has many drawbacks. Manual inspection is inefficient and easily affected by subjective factors, making it difficult to accurately identify various infectious substances, especially for some substances that look similar but have different levels of danger, which are prone to omissions. Although laboratory testing is more accurate, it takes a long time and cannot meet the needs of rapid customs clearance at ports, seriously affecting logistics efficiency and personnel flow.

[0003] Under this circumstance, it is urgent to invent an entry-exit list management system that can identify the categories of infectious substances through images. It uses advanced image acquisition and processing technology, deep learning algorithms, etc. to solve the problem of rapid and accurate identification and efficient management of infectious substances. By collecting high-definition images in key inspection areas of ports, after pre-processing such as denoising and normalization, CNN is combined with attention mechanism and manual feature fusion to extract features, and then the improved ResNet model is used to classify and divide the danger level, realizing the automatic and intelligent identification of infectious substances. With the help of functional modules such as list management, response measures execution, early warning notification and system management, the port's control ability of infectious substances is effectively improved, ensuring public health safety and entry-exit order. Summary of the invention

[0004] Technical issues solved

[0005] In view of the deficiencies of the prior art, the present invention provides an entry-exit list management system that recognizes the categories of infectious substances through images.

[0006] Technical Solution

[0007] To achieve the above-mentioned solution, the present invention provides the following technical solution: an entry-exit list management system for identifying the category of infectious substances through images, comprising the following modules:

[0008] S1 image recognition and classification module: Receive images uploaded by the front end, bilateral filtering denoising: Use the bilateral filtering formula to remove noise and retain image details. After normalization, the image data is more regular in numerical terms. Perform preliminary feature extraction on the input image through convolution operation, perform global average pooling on the feature map output by the convolution layer, and then learn the channel descriptor z through two fully connected layers to obtain the weight S of each channel. Finally, multiply the obtained weight S by the original feature map U channel by channel to obtain the feature map adjusted by the attention mechanism.

[0009] V i,j,c =U i,j,c ×S c ,

[0010] In the formula, i and j represent coordinates, c represents the channel index, and then the features extracted by CNN and the attention mechanism are fused with the manual features, and then the local binary pattern LBP features are calculated. Finally, the features extracted by CNN and the attention mechanism are concatenated with the color histogram features and LBP features calculated manually to form the final feature vector. The feature vector obtained after feature extraction and fusion is input into the improved ResNet model for processing. The improved ResNet model finally passes through two fully connected layers, and the second fully connected layer outputs the probability distribution of infectious substance categories and hazard levels through the Softmax activation function;

[0011] S2 Directory Management Module: used for data entry, data update and data query;

[0012] S3 Response Execution Module: Automatically associate corresponding response measures according to the classification and hazard level of infectious substances;

[0013] S4 Warning and Notification Module: When high-risk or medium-risk infectious substances are detected, the system automatically sends warning notifications to relevant departments and personnel;

[0014] S5 system management module: responsible for system user management, system parameter settings, and data backup and recovery.

[0015] Preferably, in the improved ResNet model, each residual block is added with a batch normalization layer, which is located after the convolution layer and before the activation function. The calculation formula of batch normalization is:

[0016]

[0017] Where x is the input of the batch normalization layer, y is the output after batch normalization, and the data distribution after processing is more stable. B is the mean of the current batch of data, is the variance of the current batch of data, which measures the degree of discreteness of the data, and the value is 10 -5 , γ and β are learnable parameters.

[0018] Preferably, the category probability vector output by the model is [P 类别1 ,P 类别2 ,…,P 类别n ], the category with the highest probability is taken as the classification result.

[0019] Preferably, for the hazard level classification task, after determining the category of the infectious substance, the probability vectors of high, medium and low hazard levels corresponding to the category [P 高 ,P 中 ,P 低 ], using the following judgment rules: P 高 ≥0.6, the infectious substance is judged to be of high risk level, P 高 <0.6 and P 中 ≥0.4, it is considered as medium risk level, P 中 If it is less than 0.4, it is judged as a low risk level.

[0020] Preferably, the response measures for high-risk infectious substances include strict control and safe destruction of related items, deep cleaning and disinfection of the premises, and comprehensive assessment of the health of relevant personnel and long-term monitoring.

[0021] Preferably, the response measures for low-risk infectious substances include, for viruses, health reminders for relevant personnel, routine inspection of goods, and normal entry and exit allowed if no abnormalities are found; for bacteria, simple disinfection of goods, basic inquiries about the health of personnel, and release if no abnormalities are found; for fungi, routine disinfection of items, inspection of personnel's skin, and entry and exit allowed if no obvious symptoms are found; for other categories, routine inspection of goods and personnel, and release after passing the inspection.

[0022] Beneficial Effects

[0023] Compared with the prior art, the present invention provides an entry-exit list management system that uses images to identify the categories of infectious substances, which has the following beneficial effects:

[0024] 1. This entry-exit list management system uses image recognition to identify the category of infectious substances. In terms of image recognition, the algorithm integrates CNN and attention mechanism and combines manual features to extract image features more accurately, thereby accurately classifying infectious substances and determining the danger level, reducing misjudgments, and effectively improving the accuracy and reliability of recognition. The response measure execution module is linked with the early warning notification module to ensure the timely implementation of prevention and control measures, and to provide efficient early warnings to relevant departments, enhancing emergency response capabilities and coordination. The overall architecture and process design are closely coordinated between the front and back ends, and the work process forms a closed loop, which not only improves the operational convenience of port staff, but also provides strong data support for management decisions, effectively ensuring the safety and efficient management of ports, and showing outstanding advantages in the field of infectious substance control. BRIEF DESCRIPTION OF THE DRAWINGS

[0025] Figure 1 It is a schematic diagram of the system function modules of the present invention;

[0026] Figure 2 Schematic diagram of the system workflow of the present invention. DETAILED DESCRIPTION

[0027] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. 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 creative work are within the scope of protection of the present invention.

[0028] See also Figure 1-2 The present invention proposes an entry-exit list management system for identifying infectious substance categories through images, including the following contents:

[0029] 1. System Architecture

[0030] Front-end interface: Design user interface to facilitate port staff operations, including login interface, which requires user name and password for identity authentication; image upload interface, which supports uploading multiple image formats and provides image preview function; result display interface, which displays the classification, hazard level, response measures and other information of infectious substances in a clear table form, and provides printing and export functions. A data query interface is also set up to query entry and exit records by date, category, hazard level and other conditions.

[0031] Backend system: adopts server-database architecture. The server is responsible for running the image recognition algorithm, processing business logic and interacting with the front end. The database is used to store information such as image samples, recognition results, entry and exit records, and response measures of infectious substances. The database design includes an infectious substance information table, which records the category, name, and hazard level of infectious substances; an image record table, which stores relevant information of uploaded images; a recognition result table, which records the results of image recognition; and an entry and exit record table, which records the entry and exit information of goods or personnel and the corresponding processing.

[0032] 2. System Function Module

[0033] S1 Image Recognition and Classification Module: Receives images uploaded by the front end, and uses image recognition algorithms to classify infectious substances and divide them into dangerous levels. The recognition results are stored in the database and fed back to the front end for display.

[0034] (1) Data input and preprocessing

[0035] 1. Image acquisition and input: Use high-definition and stable image acquisition equipment to accurately collect image data containing infectious substances in key inspection areas at entry and exit ports, such as cargo security inspection channels and passenger baggage inspection points. Ensure that the collected images have high resolution, good color reproduction and clear details, providing a high-quality data foundation for subsequent processing.

[0036] 2. Bilateral filtering denoising: Use the bilateral filtering formula

[0037]

[0038] To remove noise and preserve image details. Where f(i,j) is the pixel value of the original image at position (i,j), and g(i,j) is the pixel value of the filtered image at position (i,j). The weight function w(i,j,m,n) is composed of the spatial proximity weight w d and pixel similarity weight w r Multiply them together to get w(i,j,m,n)=w d × r ,

[0039]

[0040] Where m and n represent the relative coordinates of the neighborhood pixels centered at (i, j), and w d reflects the influence of the spatial distance between pixels on the weight, σ d is the standard deviation of spatial distance, which is used to control the rate at which the spatial proximity weight changes with distance. The pixel similarity weight w r The calculation formula is

[0041]

[0042] w r It reflects the effect of pixel value difference on weight, σ r Control the influence of pixel value differences by reasonably adjusting σ d and σ r These two key parameters can effectively remove image noise while retaining important detail information such as image edges and textures to the greatest extent, laying a good foundation for subsequent feature extraction work.

[0043] 3. Normalization: In order to make the image data meet the input requirements of the model and improve the training efficiency and accuracy of the model, the normalization method is used to normalize the pixel values ​​of the image to the [1,1] interval. The specific calculation formula is: Where x represents the original pixel value, x min and x max are the minimum and maximum pixel values ​​of the channel, respectively. After normalization, the image data is more regular in value, which helps the model better learn and identify features in the image.

[0044] (2) Feature extraction

[0045] 1. Combination of CNN and attention mechanism:

[0046] First, we use the powerful feature extraction capability of convolutional neural networks to perform convolution operations.

[0047]

[0048] Perform preliminary feature extraction on the input image. X is the input feature map, Y is the feature map output after convolution, K is the convolution kernel, (i, j) is the coordinate of the output feature map, and (m, n) is the coordinate of the convolution kernel. In this process, the convolution kernel K slides on the input feature map X, and extracts local features in the image, such as edges, corners, etc., through convolution operations.

[0049] Then, the SENet attention mechanism is introduced to further optimize the feature extraction process. The specific steps are as follows:

[0050] Perform global average pooling on the feature map output by the convolutional layer, using the formula

[0051]

[0052] Where U is the feature map output by the convolutional layer, H and W are the height and width of the feature map, respectively, and c represents the channel index. Each feature map is compressed into a value to obtain the channel descriptor z. This operation can effectively capture the global information of the entire feature map in each channel.

[0053] The channel descriptor z is then learned through two fully connected layers to obtain the weight S of each channel.

[0054] The calculation formula is

[0055] S=σ(W 2 (δ(W 1 (z)))),

[0056] Where σ is the Sigmoid function, which maps the input value to the interval (0, 1) to obtain the weight coefficient of each channel, which is used to measure the importance of the channel; δ is the ReLU function, which is used as an activation function to introduce nonlinear factors into the model and enhance the expressiveness of the model; W 1 and W 2 is the weight matrix of the two fully connected layers, which is used to linearly transform the input data; z is the channel descriptor obtained previously.

[0057] Finally, the obtained weight S is multiplied by the original feature map U channel by channel to obtain the feature map adjusted by the attention mechanism.

[0058] V i,j,c =U i,j,c ×S c ,

[0059] V i,j,c It is a feature map adjusted by the attention mechanism. It improves the pertinence and effectiveness of feature extraction by emphasizing the features of important channels and suppressing the features of unimportant channels. It can pay more attention to channel features that are important for classification and hazard level division, thereby improving the effectiveness and accuracy of feature extraction.

[0060] 2. Fusion of manual features:

[0061] In order to further enrich the expressive power of feature vectors, the features extracted by CNN and attention mechanism are fused with manual features. First, the color histogram features are calculated and the image is converted from RGB color space to HSV color space, because HSV color space is more consistent with the way humans perceive color. Then the histograms of the three channels H, S, and V are calculated respectively. The histogram of each channel is divided into 32 bins. Finally, the histograms of these three channels are spliced ​​into a feature vector. This can effectively describe the color distribution characteristics of the image.

[0062] Next, the local binary pattern (LBP) feature is calculated. A 3×3 neighborhood is taken around each pixel in the image, and the pixel value in the neighborhood is compared with the central pixel value. If the neighborhood pixel value is greater than or equal to the central pixel value, it is recorded as 1; otherwise, it is recorded as 0. In this way, an 8-bit binary number can be obtained, which is converted into a decimal number as the LBP value of the pixel. Then the frequency of occurrence of different LBP values ​​in the image is counted to form an LBP feature histogram. The LBP feature can well describe the texture information of the image.

[0063] Finally, the features extracted by CNN and the attention mechanism are concatenated with the color histogram features and LBP features calculated manually to form the final feature vector.

[0064] (3) Classification and hazard level classification

[0065] 1. Processing through the ResNet model: The feature vector obtained after feature extraction and fusion is input into the improved ResNet model for processing. In the improved ResNet model, each residual block adds a batch normalization layer, which is located after the convolution layer and before the activation function. The calculation formula of batch normalization is

[0066]

[0067] Where x is the input of the batch normalization layer, usually the output of the convolutional layer; y is the output after batch normalization, and the data distribution after processing is more stable; μ B It is the mean of the current batch of data, reflecting the average level of the current batch of data; is the variance of the current batch of data, which measures the degree of discreteness of the data; the value of ò is 10 -5 , used to prevent the denominator from being 0; γ and β are learnable parameters. Batch normalization can accelerate the training convergence speed of the model and reduce the sensitivity of the model to the initial parameters. It also has a certain regularization effect and can effectively prevent the model from overfitting.

[0068] 2. Model output and judgment:

[0069] The ResNet model finally passes through two fully connected layers, and the second fully connected layer outputs the probability distribution of infectious substance categories and hazard levels through the Softmax activation function. The Softmax function converts the output of the neural network into a probability distribution, so that the sum of the probabilities of each category or hazard level is 1, which facilitates classification and level determination.

[0070] For the classification task, assume that the class probability vector output by the model is [P 类别1 ,P 类别2 ,…,P 类别n], then the category with the highest probability is taken as the classification result. That is, if P 类别k =max{P 类别1 ,P 类别2 ,…,P 类别n}, then the infectious substance is judged to belong to category k.

[0071] For the hazard level classification task, after determining the category of infectious substances, the probability vectors of high, medium and low hazard levels corresponding to the category [P 高 ,P 中 ,P 低 ], using the following judgment rules:

[0072] If P 高 If the value is ≥0.6, the infectious substance is judged to be of high risk level. This means that the model believes that the infectious substance has a high risk and strict prevention and control measures are required.

[0073] If P 高 <0.6 and P 中 If the value is ≥0.4, it is considered as medium risk level. At this time, it is necessary to conduct further inspection and monitoring of relevant items or personnel, and take appropriate measures according to the specific situation.

[0074] If P 中 If the value is less than 0.4, it is judged as a low risk level. In this case, although the risk of infectious substances is relatively low, necessary routine inspections and records are still required.

[0075] S2 Directory Management Module:

[0076] Data entry: Staff can manually enter relevant information about infectious substances, such as newly added categories and names of infectious substances, to supplement detailed information that cannot be obtained through image recognition.

[0077] Data update: When there is new infectious substance information, identification criteria or changes in response measures, the data in the database can be updated in a timely manner.

[0078] Data query: supports multiple query methods, such as querying the entry and exit records of infectious substances within a certain period of time by date range; querying the entry and exit of specific categories of infectious substances by category; querying the relevant information of infectious substances of different hazard levels by hazard level.

[0079] S3 Countermeasures Execution Module: Automatically associate corresponding countermeasures according to the classification and hazard level of infectious substances. Staff can view the details of the countermeasures in the system and record the implementation of the measures, such as disinfection time, quarantined personnel information, etc. The system will remind you of countermeasures that have not been implemented or have been incompletely implemented.

[0080] S4 Warning and Notification Module: When high-risk or medium-risk infectious substances are detected, the system automatically sends warning notifications to relevant departments and personnel, such as the heads of health and epidemic prevention departments and port management departments. Notification methods can include text messages, emails, etc. At the same time, a striking warning prompt is displayed on the system interface to remind staff to handle it in time.

[0081] S5 system management module: responsible for system user management, including user addition, deletion, permission setting, etc.; system parameter setting, such as parameter adjustment of image recognition algorithm, warning threshold setting, etc.; data backup and recovery, regular database backup to prevent data loss, and data recovery when needed.

[0082] 3. System Workflow

[0083] (1) Image upload and recognition: Port staff upload the image of the infectious substance to be tested on the front-end interface. After the system back-end receives the image, it calls the image recognition and classification module for processing to identify the category and hazard level of the infectious substance.

[0084] (2) Result storage and display: The identification results are stored in the database and displayed on the front-end result display interface, including the classification of infectious substances, hazard level and corresponding response measures.

[0085] (3) Implementation of response measures: The staff will handle the relevant goods or personnel according to the response measures displayed by the system and record the implementation status in the system.

[0086] 1. Countermeasures for high-risk infectious substances

[0087] Viruses: such as Ebola virus, smallpox virus, etc., once detected, the highest level of emergency response mechanism will be immediately activated.

[0088] Bacteria: For example, Bacillus anthracis, etc., the goods and transportation vehicles involved should be thoroughly disinfected and rendered harmless, such as by high temperature, high pressure sterilization or chemical disinfection. Emergency medical examinations and preventive treatments should be conducted on those who have been in contact with the disease to prevent the spread of the disease.

[0089] Fungi: such as Cryptococcus neoformans (strains that cause central nervous system infections), etc., strictly disinfect or destroy related items, and use appropriate disinfectants to comprehensively disinfect the environment that may be contaminated.

[0090] Other categories: such as prions, all related items are strictly controlled, entry and exit are prohibited, and they are safely destroyed. The places involved are deeply cleaned and disinfected, and special disinfection methods are used to ensure that prions are inactivated. Comprehensive health assessments and long-term medical monitoring are carried out on relevant personnel.

[0091] 2. Countermeasures for medium-risk infectious substances

[0092] Viruses: For example, influenza virus (certain highly pathogenic subtypes), dengue virus, etc., the relevant goods or personnel are temporarily isolated and observed. The general observation period is determined according to the incubation period of the virus, such as 1-3 days for influenza virus and 5-8 days for dengue virus. Carry out multiple sampling tests on goods to ensure their safety. Conduct health checks on personnel, such as temperature testing, symptom inquiries, etc. If the test results are normal, they can be allowed to enter and exit the country after taking certain protective measures; if the test results are abnormal, further diagnosis and treatment will be carried out immediately.

[0093] Bacteria: such as Staphylococcus aureus (resistant strains), Vibrio cholerae, etc., the goods are strictly disinfected, which can be done by soaking in chemical disinfectants, fumigation, etc. Medical examinations are conducted on relevant personnel, such as collecting samples for bacterial culture testing. According to the examination results, people infected or carrying bacteria are isolated and treated, and close contacts are placed under medical observation. After confirming safety, goods and other personnel are allowed to enter and exit the country.

[0094] Fungi: such as Candida albicans (strains that cause serious mucosal infections), etc. Disinfect related items, such as soaking or wiping with antifungal drugs. Check the oral cavity, skin and other parts of the personnel, and if symptoms of infection are found, treat them in time. After observing the items and personnel for a period of time, decide whether to allow entry and exit based on the situation.

[0095] Other categories: For example, for malarial parasites, detailed medical examinations will be conducted on relevant personnel, including blood tests, to determine whether they are infected. Items that may contain mosquitoes carrying malarial parasites will be strictly inspected and handled, such as spraying with insecticides. Infected persons will receive standardized treatment, and after recovery, they will be allowed to enter and leave the country.

[0096] 3. Countermeasures for low-risk infectious substances

[0097] Viruses: such as some common cold viruses, etc. Health tips are given to relevant personnel, telling them to pay attention to personal hygiene and protection, such as wearing masks and washing hands frequently. Routine inspections are carried out on goods, including appearance inspections and packaging integrity inspections. After confirming that there are no abnormalities, they are allowed to enter and exit normally.

[0098] Bacteria: For non-pathogenic or conditionally pathogenic bacteria such as some lactic acid bacteria, simple disinfection treatment is carried out on the goods, such as surface wiping disinfection. Basic health inquiries are conducted on personnel, and if there are no abnormalities, they are allowed to enter and exit.

[0099] Fungi: For example, some common dermatophytes, etc., the relevant items shall be routinely disinfected, such as ultraviolet irradiation disinfection. The personnel shall be examined for skin, and if no obvious symptoms of infection are found, they shall be allowed to enter and exit normally.

[0100] Other categories: some common parasites with less pathogenicity, etc., routine inspections are carried out on goods and personnel, such as random inspections of goods and body surface inspections of personnel, etc. After passing the inspection, they are allowed to enter and exit the country.

[0101] (4) Data query and statistics: Staff can query the relevant records of entry and exit of infectious substances through the list management module and conduct data statistical analysis, such as counting the entry and exit quantities of various infectious substances in different time periods, to provide data support for management decisions.

[0102] (5) System maintenance and update: The system administrator maintains and updates the system through the system management module, including user management, parameter settings, data backup and other operations to ensure the normal operation of the system and data security.

[0103] Although 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 the embodiments without departing from the principles and spirit of the present invention, and that the scope of the present invention is defined by the appended claims and their equivalents.

Claims

1. An entry-exit list management system that uses images to identify the categories of infectious substances, characterized by: Includes the following modules: S1 image recognition and classification module: Receive images uploaded by the front end, bilateral filtering denoising: Use the bilateral filtering formula to remove noise and retain image details. After normalization, the image data is more regular in numerical terms. Perform preliminary feature extraction on the input image through convolution operation, perform global average pooling on the feature map output by the convolution layer, and then learn the channel descriptor z through two fully connected layers to obtain the weight S of each channel. Finally, multiply the obtained weight S by the original feature map U channel by channel to obtain the feature map adjusted by the attention mechanism. V i,j,c =U i,j,c ×S c , In the formula, i and j represent coordinates, c represents the channel index, and then the features extracted by CNN and the attention mechanism are fused with the manual features, and then the local binary pattern LBP features are calculated. Finally, the features extracted by CNN and the attention mechanism are concatenated with the color histogram features and LBP features calculated manually to form the final feature vector. The feature vector obtained after feature extraction and fusion is input into the improved ResNet model for processing. The improved ResNet model finally passes through two fully connected layers, and the second fully connected layer outputs the probability distribution of infectious substance categories and hazard levels through the Softmax activation function; S2 Directory Management Module: used for data entry, data update and data query; S3 Response Execution Module: Automatically associate corresponding response measures according to the classification and hazard level of infectious substances; S4 Warning and Notification Module: When high-risk or medium-risk infectious substances are detected, the system automatically sends warning notifications to relevant departments and personnel; S5 system management module: responsible for system user management, system parameter settings, and data backup and recovery.

2. According to claim 1, the entry-exit list management system for identifying infectious substances by image recognition is characterized by: In the improved ResNet model, each residual block adds a batch normalization layer, which is located after the convolution layer and before the activation function. The calculation formula of batch normalization is: Where x is the input of the batch normalization layer, y is the output after batch normalization, and the data distribution after processing is more stable. B is the mean of the current batch of data, is the variance of the current batch of data, which measures the degree of discreteness of the data, and the value is 10 -5 , γ and β are learnable parameters.

3. The entry-exit list management system for identifying infectious substances by image recognition according to claim 1, characterized in that: The class probability vector output by the model is [P 类别1 ,P 类别2 ,…,P 类别n ], the category with the highest probability is taken as the classification result.

4. The entry-exit list management system for identifying infectious substances by image recognition according to claim 3, characterized in that: For the hazard level classification task, after determining the category of infectious substances, the high, medium and low hazard level probability vectors [P 高 ,P 中 ,P 低 ], using the following judgment rules: P 高 ≥0.6, the infectious substance is judged to be of high risk level, P 高 <0.6 and P 中 ≥0.4, it is considered as medium risk level, P 中 If it is less than 0.4, it is judged as a low risk level.

5. The entry-exit list management system for identifying infectious substances by image according to claim 4, characterized in that: The above-mentioned response measures for high-risk infectious substances include strict control and safe destruction of related items, deep cleaning and disinfection of premises, and comprehensive assessment of the health of relevant personnel and long-term monitoring.

6. The entry-exit list management system for identifying infectious substances by image recognition according to claim 4, characterized in that: The response measures for low-risk infectious substances include, for viruses, health reminders for relevant personnel, routine inspection of goods, and normal entry and exit allowed if no abnormalities are found; for bacteria, simple disinfection of goods, basic inquiries about the health of personnel, and release if no abnormalities are found; for fungi, routine disinfection of items, inspection of personnel's skin, and entry and exit allowed if no obvious symptoms are found; for other categories, routine inspection of goods and personnel, and release after passing the inspection.

Images