Antigen detection kit automatic identification method and system based on pp-yoloe

Abstract

The application provides an antigen detection kit automatic identification method and system based on PP-YOLOE. The method comprises the following steps: 1) collecting a large amount of antigen detection kit image data to form a known data set Z; 2) establishing a feature vector database D of negative / positive samples; 3) using a PP-YOLOE model trained on the data set Z to perform target detection, obtaining a "CT" region ROI on the antigen detection kit, and simultaneously predicting a detection result P1; 4) if P1 is located between a negative threshold and a positive threshold, a trained classification model is used to extract features of the ROI region, and then a search is performed in the already constructed feature vector database to find the feature vector with the highest similarity, that is, it is considered that the sample label is consistent with the sample label in the database, and an identification result is obtained. Compared with the prior art, the application has better generalization, more effective execution speed and higher recognition accuracy.

Description

Technical Field

[0001] This application relates to the field of computer vision application technology, and in particular to an automatic identification method and system for antigen detection kits based on PP-YOLOE. Background Technology

[0002] One-stage object detectors are popular in real-time applications due to their speed and accuracy advantages. The most prominent architecture among single-stage detectors is the YOLO family. Since YOLOv1, the YOLO family of object detectors has undergone significant changes in network structure and label assignment. Currently, YOLOX achieves the best balance between speed and accuracy at 50.1 mAP at 68.9 FPS on a Tesla V100. YOLOX introduces an advanced anchorless method with dynamic label assignment to improve detector performance, significantly outperforming YOLOv5 in accuracy. Inspired by YOLOX, PP-YOLOv2 is a high-performance single-stage detector with 49.5 mAP at 68.9 FPS on a Tesla V100. Based on PP-YOLOv2, PP-YOLOE avoids the use of operators such as deformable convolution and matrix NMS, thus achieving good support on various hardware. Furthermore, PPYOLOE can be easily scaled to a range of models on various hardware with different computing capabilities. These characteristics further promote the application of PP-YOLOE in a wider range of real-world scenarios.

[0003] Currently, there is no technical solution for using PP-YOLOE in antigen kit detection. Summary of the Invention

[0004] In view of this, this invention, taking into account the characteristics of antigen detection kit images and adhering to the principle of "rather falsely detect negative samples than miss any positive samples," proposes an automatic identification method and system for antigen detection kits based on PP-YOLOE. This application can specifically solve the existing problems.

[0005] To achieve the above objectives, this application proposes an automated identification method for an antigen detection kit based on PP-YOLOE, comprising:

[0006] 1) Collect a large amount of image data of antigen detection kits to form a known dataset;

[0007] 2) Establish a feature vector database for negative / positive samples;

[0008] 3) Use the PP-YOLOE model trained on the known dataset to perform target detection on the target antigen detection kit image, obtain the ROI region on the target antigen detection kit, and simultaneously predict the detection result P1;

[0009] 4) If P1 is between the negative threshold and the positive threshold, use the trained classification model to extract features from the ROI region. Then, search in the constructed feature vector database to find the feature vector with the highest similarity. The label that the ROI region should be assigned is called the sample label. The label of the feature vector with the highest similarity is called the sample label in the database. That is, the sample label is considered to be consistent with the sample label in the database, and the recognition result is obtained.

[0010] Furthermore, in step 1), the collected antigen test kit image data contains several test kits in a single image.

[0011] Furthermore, in step 2), establishing the feature vector database includes the following steps:

[0012] The SIFT algorithm is used to extract the feature point set of the original image; the MeanShift algorithm is used to cluster the feature point set.

[0013] Store in the database.

[0014] Furthermore, the implementation details of the SIFT algorithm are as follows:

[0015] L(x,y,σ)=G(x,y,σ)*I(x,y)

[0016]

[0017] Where L(x, y, σ) is defined as the convolution operation between the original image I(x, y) and a variable-scale 2D Gaussian function G(x, y, σ);

[0018] G(x, y, σ) is a scale-variable Gaussian function, (x, y) are spatial coordinates, and σ represents the smoothness of the image;

[0019] Find the extreme points in the Gaussian difference scale space. If a point is the maximum or minimum value in the 26 neighborhoods of the current layer and the two layers above and below in the Gaussian difference scale space, then the point is considered a feature point of the image at that Gaussian difference scale.

[0020] Furthermore, in step 3), PP-YOLOE is trained on the known dataset, and then the ROI region and the recognition result P1 of the ROI region are exported from the input image.

[0021] Furthermore, step 4) specifically includes the following steps:

[0022] Curve fitting is performed on the DOG function in the difference-of-Gaussian scale space to generate descriptors for two feature points. The descriptors of each order in the two feature points are matched, and a match of 128 dimensions indicates that the two feature points are matched.

[0023] Furthermore, in step 4),

[0024] If the predicted result is not between the negative and positive thresholds, then the result is directly determined as negative or positive.

[0025] To achieve the above objectives, this application also proposes an automated identification system for antigen detection kits based on PP-YOLOE, comprising:

[0026] The data collection module is used to collect a large amount of image data from antigen test kits to form a known dataset;

[0027] The database creation module is used to create a feature vector database for negative / positive samples;

[0028] The target detection module is used to perform target detection on the target antigen test kit image using the PP-YOLOE model trained on the known dataset, obtain the ROI region on the target antigen test kit, and simultaneously predict the detection result P1.

[0029] The identification module extracts features from the ROI region using a trained classification model if P1 is between the negative and positive thresholds. Then, it searches the pre-built feature vector database to find the feature vector with the highest similarity. This indicates that the sample label of the target antigen detection kit matches the sample label in the database, thus obtaining the identification result.

[0030] In summary, the advantages of this application and the user experience it brings are as follows:

[0031] First, by combining advanced computer vision technologies, including object detection and feature matching, the efficiency is greatly improved compared to manual deduplication.

[0032] Second, the model of this invention has better generalization ability and can adapt to various uneven lighting environments;

[0033] Third, this invention can quickly screen antigen results, and to a large extent eliminates the phenomena of concealment, false reporting, and falsification of antigen test results. Attached Figure Description

[0034] In the accompanying drawings, unless otherwise specified, the same reference numerals throughout the various drawings denote the same or similar parts or elements. These drawings are not necessarily drawn to scale. It should be understood that these drawings depict only some embodiments disclosed in this application and should not be construed as limiting the scope of this application.

[0035] Figure 1 This is a flowchart illustrating the calculation process of the method of the present invention.

[0036] Figure 2 This is a network structure diagram of the method of the present invention.

[0037] Figure 3 This is a sample image of the detection method of the present invention.

[0038] Figure 4 A schematic diagram of the automatic identification system for an antigen detection kit based on PP-YOLOE according to an embodiment of this application is shown.

[0039] Figure 5 A schematic diagram of the structure of an electronic device provided in one embodiment of this application is shown.

[0040] Figure 6 A schematic diagram of a storage medium provided in one embodiment of this application is shown. Detailed Implementation

[0041] The present application will now be described in further detail with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are for illustrative purposes only and are not intended to limit the invention. Furthermore, it should be noted that, for ease of description, only the parts relevant to the invention are shown in the accompanying drawings.

[0042] It should be noted that, unless otherwise specified, the embodiments and features described in this application can be combined with each other. This application will now be described in detail with reference to the accompanying drawings and embodiments.

[0043] Step 1: Collect a large amount of antigen test kit image data from community and school epidemic management departments to form a known dataset Z, which is stored in the OPENML machine learning database.

[0044] Step 2: Use the SIFT algorithm to extract features from the original image, establish a feature vector database D for negative / positive samples, and use the MeanShift algorithm to cluster the feature vectors to improve data purity.

[0045] Step 3: Use the PP-YOLOE model trained on dataset Z to perform target detection, obtain the “CT” region ROI on the antigen test kit, and simultaneously predict the detection result P1.

[0046] Step 4): If P1 is between the two thresholds negative_threshed and positive_threshed, use the trained classification model to extract features from the ROI region. Then, search in the constructed feature vector database D to find the feature vector with the highest similarity. The label that should be assigned to this ROI region is called the sample label, and the label of the found feature vector with the highest similarity is called the sample label in the database. That is, it is considered that the sample label is consistent with the sample label in the database, and the recognition result is obtained. Figure 3 As shown.

[0047] In step 1), the collected antigen reagent kit image data can contain several reagent kits in a single image.

[0048] Step 2) specifically includes the following steps:

[0049] 5) Use the SIFT algorithm to extract the feature point set of the original image. The details of the SIFT algorithm implementation are as follows:

[0050] L(x,y,σ)=G(x,y,σ)*I(x,y)

[0051]

[0052] This is an initialization operation. The purpose of scale space theory is to simulate the multi-scale features of image data. The Gaussian kernel is the only kernel that can generate multi-scale space. The scale space of an image, L(x, y, σ), is defined as the convolution operation between the original image I(x, y) and a variable-scale 2D Gaussian function G(x, y, σ).

[0053] Where G(x, y, σ) is a scale-variable Gaussian function, (x, y) are spatial coordinates, and σ is the scale coordinate. The magnitude of σ determines the smoothness of the image; a large scale corresponds to the overall features of the image, while a small scale corresponds to the detailed features. A large σ value corresponds to a coarse scale (low resolution), and vice versa. To effectively detect stable keypoints in the scale space, this invention proposes a Difference-of-Gaussian (DOG) scale-space. This is generated by convolving the image with Gaussian difference kernels of different scales.

[0054] D(x,y,σ)=[G(x,y,kσ)-G(x,y,σ)]*O(k,y)=L(x,y,kσ)-L(x,y,σ)

[0055] Where k is an empirical constant.

[0056] To find extrema in scale space, each sampled point is compared with all its neighbors to see if it is larger or smaller than its neighbors in both the image domain and scale domain. The intermediate detection point is compared with its 8 neighbors at the same scale and 9 × 2 points corresponding to the adjacent scales above and below, a total of 26 points, to ensure that extrema are detected in both scale space and the two-dimensional image space. A point is considered a feature point of the image at that scale if it is the maximum or minimum value in its 26 neighborhoods within the current layer and the two layers above and below in the DOG scale space.

[0057] 6) Store in the database, using the following database statements:

[0058] CREATE DATABASE ANTIGEN_KIT;

[0059] USE ANTIGEN_KIT;

[0060] CREATE TABLE existing_antigen_kit(antigen_kit_id VARCHAR(32),img_pathVARCHAR(64));

[0061] INSERT INTO existing_antigen_kit VALUES("***","***");

[0062] Step 3) specifically includes the following steps:

[0063] 7) The overall architecture of PP-YOLOv2 consists of a ResNet50vd backbone with deformable convolutions, a PAN neck with SPP layers and DropBlock, and a lightweight IoU-aware head. In PPYOLOv2, the backbone uses ReLU activation, while the neck uses mish activation. Following YOLOv3, PP-YOLOv2 only assigns one anchor box per ground truth object. In addition to classification loss, regression loss, and object loss, PP-YOLOv2 also uses IoU loss and IoU-aware loss to improve performance.

[0064] 8) As mentioned above, PP-YOLOv2 assigns ground truth (GT) in an anchor-based manner. However, the anchor mechanism introduces many hyperparameters and relies on hand-design, which may not generalize well on other datasets. An anchor-free method is introduced in PP-YOLOv2. After FCOS (tiling an anchor point on each pixel), three detector heads set upper and lower bounds to assign GT to the corresponding feature maps. Then, the center of the bounding box is calculated to select the nearest pixel as the positive sample. After the YOLO series, a predicted 4D vector (x, y, w, h) is used for regression.

[0065] 9) Regarding the task alignment loss L cls-ρos It uses the maximum IoU for each instance as the standardization. The binary cross-entropy (BCE) for classification can be rewritten as:

[0066]

[0067] Where p i t represents the prediction result for the i-th sample. i This represents the true value of the i-th sample.

[0068] 10) For classification and localization tasks, the variable focal length loss (VFL) is selected respectively. VFL and Distributed Focal Length Loss (DFL) DFL PP Picodet successfully applied VFL and DFL to object detectors and achieved performance improvements. For VFL, unlike Quality Focus Loss (QFL), VFL uses the target score to weight the loss for positive samples. This implementation makes positive samples with high IoU contribute relatively more to the loss. This also makes the model focus more on high-quality samples rather than low-quality samples during training. Similarly, both use the IoU-aware classification score (IACS) as the prediction target. This effectively learns a joint representation of the classification score and localization quality estimate, achieving high consistency between training and inference. For DFL, to address the inflexible bounding box representation, a general distribution is used to predict the bounding boxes. The model is supervised by the loss function:

[0069]

[0070] a, β, and γ are empirical constants, t is the true value of the sample, and loss GIOU Γ represents the GIoU loss, and N is the number of samples.

[0071] Step 4) specifically includes the following steps:

[0072] 11) If P1 lies between the negative_threshed and positive_threshed thresholds, the trained classification model is used to extract features from the ROI region. Then, a search is performed in the pre-built feature vector database to find the feature vector with the highest similarity. This indicates that the sample label matches the sample label in the database, resulting in the identification result. The negative_threshed value is 0.2, and the positive_threshed value is 0.7. Since DoG values ​​are sensitive to noise and edges, further verification is needed to accurately locate feature points after detecting local extrema in the DoG scale space. To improve the stability of feature points, curve fitting of the scale-space DoG function is required. This is achieved using the Taylor expansion of the DoG function in scale space.

[0073]

[0074] Where X = (x, y, σ) T , where x and y are coordinates, and σ is a constant.

[0075] 12) After generating feature points (128 dimensions) for the ROI region, match them with feature points of each sample in the database. A match across 128 dimensions indicates that the two feature points can be used to calculate distance. This process compares the nearest neighbor distance with the second nearest neighbor distance. A distance ratio less than a certain threshold is considered a correct match. This is because, due to the high dimensionality of the feature space, similar distances may result in numerous other incorrect matches, leading to a high ratio value. The principle for determining the ratio is as follows:

[0076] A ratio of 0.4 is suitable for matching with high accuracy requirements;

[0077] A ratio of 0.6 is suitable for matching scenarios that require a large number of matching points.

[0078] ratio = 0.5 under normal circumstances;

[0079] Alternatively, the following principle can be followed: when the nearest neighbor distance is less than 200, ratio = 0.6; otherwise, ratio = 0.4. This strategy for determining the ratio can eliminate incorrect matching points.

[0080] 13) After feature matching, read the sample labels from the database and determine whether they are negative or positive.

[0081] The application provides an automatic identification system for antigen detection kits based on PP-YOLOE. This system is used to execute the automatic identification method for antigen detection kits based on PP-YOLOE described in the above embodiments, such as... Figure 4 As shown, the system includes:

[0082] Data collection module 401 is used to collect a large amount of antigen test kit image data to form a known dataset;

[0083] Database creation module 402 is used to create a feature vector database for negative / positive samples;

[0084] The target detection module 403 is used to perform target detection on the target antigen test kit image using the PP-YOLOE model trained on the known dataset, obtain the ROI region on the target antigen test kit, and simultaneously predict the detection result P1.

[0085] The identification module 404, if P1 is located between the negative threshold and the positive threshold, uses the trained classification model to extract features from the ROI region, and then searches in the constructed feature vector database to find the feature vector with the highest similarity. This means that the sample label of the target antigen detection kit is considered to be consistent with the sample label in the database, and the identification result is obtained.

[0086] The PP-YOLOE-based automatic identification system for antigen detection kits provided in the above embodiments of this application and the PP-YOLOE-based automatic identification method for antigen detection kits provided in the embodiments of this application are based on the same inventive concept and have the same beneficial effects as the methods used, run or implemented by their stored applications.

[0087] This application also provides an electronic device corresponding to the PP-YOLOE-based automatic identification method for antigen detection kits provided in the foregoing embodiments, for executing the PP-YOLOE-based automatic identification method for antigen detection kits. This application does not limit the scope of the embodiments.

[0088] Please refer to Figure 5 This illustrates a schematic diagram of an electronic device provided by some embodiments of this application. For example... Figure 5 As shown, the electronic device 20 includes: a processor 200, a memory 201, a bus 202, and a communication interface 203. The processor 200, the communication interface 203, and the memory 201 are connected via the bus 202. The memory 201 stores a computer program that can run on the processor 200. When the processor 200 runs the computer program, it executes the automatic identification method for antigen detection kits based on PP-YOLOE provided in any of the foregoing embodiments of this application.

[0089] The memory 201 may include high-speed random access memory (RAM) or non-volatile memory, such as at least one disk storage device. Communication between this system network element and at least one other network element is achieved through at least one communication interface 203 (which can be wired or wireless), such as the Internet, wide area network, local area network, or metropolitan area network.

[0090] Bus 202 can be an ISA bus, PCI bus, or EISA bus, etc. The bus can be divided into an address bus, a data bus, a control bus, etc. The memory 201 is used to store programs. After receiving an execution instruction, the processor 200 executes the program. The automatic identification method for antigen detection kits based on PP-YOLOE disclosed in any of the foregoing embodiments of this application can be applied to the processor 200, or implemented by the processor 200.

[0091] The processor 200 may be an integrated circuit chip with signal processing capabilities. In implementation, each step of the above method can be completed by the integrated logic circuitry in the hardware of the processor 200 or by instructions in software form. The processor 200 may be a general-purpose processor, including a central processing unit (CPU), a network processor (NP), etc.; it may also be a digital signal processor (DSP), an application-specific integrated circuit (ASIC), an off-the-shelf programmable gate array (FPGA), or other programmable logic devices, discrete gate or transistor logic devices, or discrete hardware components. It can implement or execute the methods, steps, and logic block diagrams disclosed in the embodiments of this application. The general-purpose processor may be a microprocessor or any conventional processor. The steps of the methods disclosed in the embodiments of this application can be directly embodied in the execution of a hardware decoding processor, or executed by a combination of hardware and software modules in the decoding processor. The software modules may reside in random access memory, flash memory, read-only memory, programmable read-only memory, electrically erasable programmable memory, registers, or other mature storage media in the art. The storage medium is located in memory 201. The processor 200 reads the information in memory 201 and, in conjunction with its hardware, completes the steps of the above method.

[0092] The electronic device provided in this application embodiment and the automatic identification method for antigen detection kit based on PP-YOLOE provided in this application embodiment are based on the same inventive concept and have the same beneficial effects as the methods they employ, operate, or implement.

[0093] This application also provides a computer-readable storage medium corresponding to the automatic identification method for antigen detection kits based on PP-YOLOE provided in the foregoing embodiments. Please refer to... Figure 6 The computer-readable storage medium shown is an optical disc 30, on which a computer program (i.e., a program product) is stored. When the computer program is run by a processor, it executes the automatic identification method for the antigen detection kit based on PP-YOLOE provided in any of the foregoing embodiments.

[0094] It should be noted that examples of the computer-readable storage medium may also include, but are not limited to, phase-change memory (PRAM), static random access memory (SRAM), dynamic random access memory (DRAM), other types of random access memory (RAM), read-only memory (ROM), electrically erasable programmable read-only memory (EEPROM), flash memory or other optical and magnetic storage media, which will not be elaborated here.

[0095] The computer-readable storage medium provided in the above embodiments of this application and the automatic identification method for antigen detection kits based on PP-YOLOE provided in the embodiments of this application are based on the same inventive concept and have the same beneficial effects as the methods adopted, run or implemented by the applications stored therein.

[0096] It should be noted that:

[0097] The algorithms and displays provided herein are not inherently related to any particular computer, virtual system, or other device. Various general-purpose systems can also be used in conjunction with the teachings herein. The required structure for constructing such systems is apparent from the above description. Furthermore, this application is not directed to any particular programming language. It should be understood that the content of this application described herein can be implemented using various programming languages, and the above description of specific languages ​​is for the purpose of disclosing the best mode of implementation of this application.

[0098] Numerous specific details are set forth in the specification provided herein. However, it will be understood that embodiments of this application may be practiced without these specific details. In some instances, well-known methods, structures, and techniques have not been shown in detail so as not to obscure the understanding of this specification.

[0099] Similarly, it should be understood that, in order to simplify this application and aid in understanding one or more of the various inventive aspects, in the above description of exemplary embodiments of this application, various features of this application are sometimes grouped together into a single embodiment, figure, or description thereof. However, this method of disclosure should not be construed as reflecting an intention that the claimed application requires more features than are expressly recited in each claim. Rather, as reflected in the following claims, inventive aspects lie in fewer than all features of a single foregoing disclosed embodiment. Therefore, the claims following the detailed description are hereby expressly incorporated into that detailed description, wherein each claim itself is a separate embodiment of this application.

[0100] Those skilled in the art will understand that modules in the device of the embodiments can be adaptively changed and placed in one or more devices different from that embodiment. Modules, units, or components in the embodiments can be combined into a single module, unit, or component, and further, they can be divided into multiple sub-modules, sub-units, or sub-components. Except where at least some of such features and / or processes or units are mutually exclusive, any combination can be used to combine all features disclosed in this specification (including the accompanying claims, abstract, and drawings) and all processes or units of any method or device so disclosed. Unless expressly stated otherwise, each feature disclosed in this specification (including the accompanying claims, abstract, and drawings) may be replaced by an alternative feature that serves the same, equivalent, or similar purpose.

[0101] Furthermore, those skilled in the art will understand that although some embodiments described herein include certain features but not others included in other embodiments, combinations of features from different embodiments are intended to be within the scope of this application and form different embodiments. For example, in the following claims, any of the claimed embodiments can be used in any combination.

[0102] The various component embodiments of this application can be implemented in hardware, or as software modules running on one or more processors, or a combination thereof. Those skilled in the art will understand that microprocessors or digital signal processors (DSPs) can be used in practice to implement some or all of the functions of some or all of the components in the virtual machine creation system according to the embodiments of this application. This application can also be implemented as a device or system program (e.g., a computer program and computer program product) for performing part or all of the methods described herein. Such an implementation of this application can be stored on a computer-readable medium, or can be in the form of one or more signals. Such signals can be downloaded from an Internet website, provided on a carrier signal, or provided in any other form.

[0103] It should be noted that the above embodiments are illustrative of this application and not restrictive, and that those skilled in the art can devise alternative embodiments without departing from the scope of the appended claims. In the claims, any reference signs placed between parentheses should not be construed as limiting the claims. The word "comprising" does not exclude the presence of elements or steps not listed in the claims. The word "a" or "an" preceding an element does not exclude the presence of a plurality of such elements. This application can be implemented by means of hardware comprising several different elements and by means of a suitably programmed computer. In the unit claims enumerating several systems, several of these systems may be embodied by the same item of hardware. The use of the words first, second, and third, etc., does not indicate any order. These words can be interpreted as names.

[0104] The above description is merely a specific embodiment of this application, but the scope of protection of this application is not limited thereto. Any person skilled in the art can easily conceive of various variations or substitutions within the technical scope disclosed in this application, and these should all be included within the scope of protection of this application. Therefore, the scope of protection of this application should be determined by the scope of the claims.

Claims

1. An automated identification method for an antigen detection kit based on PP-YOLOE, characterized in that, include: 1) Collect a large amount of image data of antigen detection kits to form a known dataset; 2) Establish a feature vector database for negative / positive samples; 3) Use the PP-YOLOE model trained on the known dataset to perform target detection on the target antigen detection kit image, obtain the ROI region on the target antigen detection kit, and simultaneously predict the detection result P1; 4) If P1 is between the negative threshold and the positive threshold, use the trained classification model to extract features from the ROI region. Then, search in the constructed feature vector database to find the feature vector with the highest similarity. The label that the ROI region should be assigned is called the sample label. The label of the feature vector with the highest similarity is called the sample label in the database. That is, the sample label is considered to be consistent with the sample label in the database, and the recognition result is obtained. In step 2), establishing the feature vector database includes the following steps: extracting the feature point set of the original image using the SIFT algorithm; clustering the feature point set using the MeanShift algorithm; and storing it in the database. The implementation details of the SIFT algorithm are as follows: L(x,y,σ)=G(x,y,σ)*I(x,y) Where L(x,y,σ) is defined as the convolution operation between the original image I(x,y) and a variable-scale 2D Gaussian function G(x,y,σ); G(x,y,σ) is a scale-variable Gaussian function, (x,y) are spatial coordinates, and σ represents the smoothness of the image; Find the extreme points in the Gaussian difference scale space. If a point is the maximum or minimum value in the 26 neighborhoods of the current layer and the two layers above and below in the Gaussian difference scale space, then the point is considered a feature point of the image at that Gaussian difference scale.

2. The method according to claim 1, characterized in that, In step 1), the collected antigen test kit image data contains several kits in a single image.

3. The method according to claim 1, characterized in that, In step 3), PP-YOLOE is trained on the known dataset, and then the ROI region and the recognition result P1 of the ROI region are exported from the input image.

4. The method according to claim 1, characterized in that, Step 4) specifically includes the following steps: Curve fitting is performed on the DOG function in the difference-of-Gaussian scale space to generate descriptors for two feature points. The descriptors of each order in the two feature points are matched, and a match of 128 dimensions indicates that the two feature points are matched.

5. The method according to claim 1, characterized in that, In step 4), If the predicted result is not between the negative and positive thresholds, then the result is directly determined as negative or positive.

6. An automated identification system for antigen detection kits based on PP-YOLOE, using the method described in any one of claims 1-5, characterized in that, include: The data collection module is used to collect a large amount of image data from antigen test kits to form a known dataset; The database creation module is used to create a feature vector database for negative / positive samples; The target detection module is used to perform target detection on the target antigen test kit image using the PP-YOLOE model trained on the known dataset, obtain the ROI region on the target antigen test kit, and simultaneously predict the detection result P1. The identification module extracts features from the ROI region using a trained classification model if P1 is between the negative and positive thresholds. Then, it searches the pre-built feature vector database to find the feature vector with the highest similarity. This indicates that the sample label of the target antigen detection kit matches the sample label in the database, thus obtaining the identification result.

7. An electronic device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that, The processor executes the computer program to implement the method as described in any one of claims 1-5.

8. A computer-readable storage medium having a computer program stored thereon, characterized in that, The program is executed by a processor to implement the method as described in any one of claims 1-5.

Images