Railway passenger transport security check forbidden and limited article intelligent identification method and device

Abstract

The invention provides a railway passenger transport security check forbidden and limited article intelligent identification method and device, and the method comprises the steps: inputting a currently collected target article image corresponding to a target article located on railway passenger transport security check equipment into a preset forbidden and limited article intelligent identification model, so as to output forbidden and limited article identification result data corresponding to the target article image, the model is obtained by training a YOLOv8 optimization model based on each article image of various scales and a label in advance, and the label is used for identifying a detection frame on the article image and an identification type identifier of a forbidden and restricted article to which the detection frame belongs; the YOLOv8 optimization model is formed by optimizing a YOLOv8 model on the basis of a SegNeXt convolution attention mechanism and a GSConv module in advance; and outputting forbidden and restricted article identification result data. According to the invention, the calculation complexity of the forbidden and restricted article intelligent identification model can be effectively reduced, and the accuracy and efficiency of the forbidden and restricted article intelligent identification by using the model can be improved.

Description

Technical Field

[0001] The present application relates to the field of data processing technology, and in particular to a method and device for intelligently identifying prohibited and restricted items for railway passenger security inspection. Background Art

[0002] With the continuous growth of railway passenger volume and the increasing public demand for safety, railway passenger security inspection is facing the challenges of high security requirements, large passenger flow processing and identification of diversified prohibited and restricted items. However, current railway security inspection still relies on manual identification, and various detection equipment has limited detection capabilities for different prohibited and restricted items. The detection effect depends on manual experience, which is prone to missed detection and false detection. Among them, the prohibited and restricted items for railway passenger security inspection refer to the items that are prohibited or restricted from being carried on board, checked in or placed in the luggage compartment in accordance with relevant laws and regulations in order to protect the safety of passengers' lives and property, maintain public order and public health during railway passenger transportation.

[0003] With the continuous development of advanced technologies such as machine learning and deep learning, artificial intelligence recognition technology has provided strong support for the intelligent upgrade of the security inspection field. It can achieve fast and accurate identification of prohibited and restricted items by adopting target detection methods such as the YOLOv8 model.

[0004] However, the existing methods of target detection such as the YOLOv8 model have the problem of high model computational complexity. If a model with lower computational complexity is directly adopted, the target detection accuracy cannot be guaranteed. Summary of the invention

[0005] In view of this, the embodiments of the present application provide a method and device for intelligently identifying prohibited and restricted items in railway passenger security inspection, so as to eliminate or improve one or more defects existing in the prior art.

[0006] One aspect of the present application provides a method for intelligently identifying prohibited and restricted items for railway passenger security inspection, comprising:

[0007] Input the target item image corresponding to the target item currently collected and located on the railway passenger security inspection equipment into the preset prohibited and restricted goods intelligent recognition model, so that the prohibited and restricted goods intelligent recognition model outputs the prohibited and restricted goods recognition result data corresponding to the target item image, wherein the prohibited and restricted goods intelligent recognition model is obtained by training the YOLOv8 optimization model based on various item images of multiple scales and the labels corresponding to each of the item images, and the labels are used to identify the detection frame on the item image and the prohibited and restricted goods recognition type identifier to which the detection frame belongs; the YOLOv8 optimization model is formed by optimizing the YOLOv8 model based on the SegNeXt convolution attention mechanism and the GSConv grouping and spatial combined convolution module in advance;

[0008] Output the prohibited and restricted goods identification result data.

[0009] In some embodiments of the present application, the step of inputting the target item image corresponding to the target item currently collected and located on the railway passenger security inspection device into a preset prohibited and restricted goods intelligent recognition model so that the prohibited and restricted goods intelligent recognition model outputs prohibited and restricted goods recognition result data corresponding to the target item image includes:

[0010] The target item image corresponding to the target item currently collected and located on the railway passenger security inspection equipment is input into the backbone network based on the SegNeXt convolutional attention mechanism in the preset prohibited and restricted goods intelligent recognition model, so that the backbone network extracts the multi-scale features of the target item image, and then the backbone network inputs the multi-scale features into the feature fusion module provided with GSConv grouping and spatial combined convolution modules in the YOLOv8 model for fusion to obtain the fusion features of the target item image, and then the feature fusion module inputs the fusion features into the head network in the YOLOv8 model, so that the head network outputs the prohibited and restricted goods recognition result data corresponding to the target item image; wherein, the prohibited and restricted goods recognition result data includes a detection frame on the target item image and a prohibited and restricted goods recognition type identifier used to indicate the detection frame.

[0011] In some embodiments of the present application, the outputting the prohibited or restricted goods identification result data includes:

[0012] If the prohibited and restricted goods recognition result data corresponding to the target item image output by the prohibited and restricted goods intelligent recognition model contains multiple detection frames, one of the multiple detection frames is selected as the target boundary frame based on the non-maximum suppression method; if the prohibited and restricted goods recognition result data corresponding to the target item image output by the prohibited and restricted goods intelligent recognition model contains one detection frame, the detection frame is used as the target boundary frame;

[0013] Adjusting the target bounding box based on a bounding box regression technique;

[0014] The target bounding box and the prohibited or restricted item identification type identifier to which the target bounding box belongs are used as target prohibited or restricted item identification result data for the target item image, and the target prohibited or restricted item identification result data is output.

[0015] In some embodiments of the present application, before inputting the currently acquired target item image located on the railway passenger security inspection device into the preset prohibited and restricted items intelligent recognition model, the method further includes:

[0016] Acquire X-ray images of the same target object located on the railway passenger security inspection device from a first viewing angle and a second viewing angle, respectively, to obtain a first X-ray image corresponding to the first viewing angle and a second X-ray image corresponding to the second viewing angle, and use the first X-ray image and the second X-ray image as target object images corresponding to the target object, respectively;

[0017] The prohibited and restricted goods intelligent identification model outputs the prohibited and restricted goods identification result data corresponding to the first X-ray image and the second X-ray image respectively;

[0018] Correspondingly, before outputting the prohibited and restricted goods identification result data, the method further includes:

[0019] If the prohibited or restricted item identification type identifier to which the target boundary box belongs indicates that the item in the target boundary box is a prohibited or restricted item, then, according to the prohibited or restricted item identification result data corresponding to each of the first X-ray image and the second X-ray image, performing prohibited or restricted item length identification on the adjusted target boundary box to obtain a corresponding prohibited or restricted item length;

[0020] Among them, the prohibited and restricted goods identification result data also includes: the length of the prohibited and restricted goods.

[0021] In some embodiments of the present application, if the prohibited and restricted goods identification type identifier to which the target bounding box belongs includes: sharp objects; then, according to the prohibited and restricted goods identification result data corresponding to each of the first X-ray image and the second X-ray image, the prohibited and restricted goods length identification is performed on the adjusted target bounding box to obtain the corresponding prohibited and restricted goods length, including:

[0022] Performing feature detection, image calibration and edge detection on the sharp object in the target boundary boxes corresponding to the first X-ray image and the second X-ray image, respectively, to determine the image contour of the sharp object;

[0023] The pixel length of the image contour of the sharp object is measured, and the pixel length is converted into an actual length based on a preset scale to obtain the prohibited item length corresponding to the sharp object.

[0024] In some embodiments of the present application, before inputting the currently acquired target item image located on the railway passenger security inspection device into the preset prohibited and restricted items intelligent recognition model, the method further includes:

[0025] Obtain a dataset consisting of images of various objects at various scales;

[0026] In addition, the backbone network of the YOLOv8 model is optimized based on SegNeXt convolutional attention, and the feature fusion module of the YOLOv8 model is optimized using GSConv grouping and spatial combined convolution modules to obtain the YOLOv8 optimized model;

[0027] The YOLOv8 optimization model is trained using the data set and a preset loss function to train the YOLOv8 optimization model into a prohibited and restricted goods intelligent recognition model for outputting prohibited and restricted goods recognition result data corresponding to the object image.

[0028] In some embodiments of the present application, the step of obtaining a data set consisting of images of objects at multiple scales includes:

[0029] Obtaining images of various objects at various scales, performing data enhancement and labeling on the various object images, so as to obtain a data set consisting of the various object images at various scales and labels corresponding to the various object images.

[0030] Another aspect of the present application provides a railway passenger security inspection prohibited and restricted goods intelligent identification device, comprising:

[0031] A model recognition module is used to input the target item image corresponding to the target item currently collected and located on the railway passenger security inspection equipment into a preset prohibited and restricted goods intelligent recognition model, so that the prohibited and restricted goods intelligent recognition model outputs prohibited and restricted goods recognition result data corresponding to the target item image, wherein the prohibited and restricted goods intelligent recognition model is obtained by training a YOLOv8 optimization model based on various item images of multiple scales and labels corresponding to each of the item images, and the label is used to identify a detection frame on the item image and a prohibited and restricted goods recognition type identifier to which the detection frame belongs; the YOLOv8 optimization model is formed by optimizing the YOLOv8 model based on the SegNeXt convolutional attention mechanism and the GSConv grouping and spatial combined convolution module in advance;

[0032] The result output module is used to output the prohibited and restricted goods identification result data.

[0033] The third aspect of the present application provides an electronic device, including a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor implements the method for intelligent identification of prohibited and restricted items for railway passenger security inspection when executing the computer program.

[0034] The fourth aspect of the present application provides a computer-readable storage medium on which a computer program is stored. When the computer program is executed by a processor, the method for intelligently identifying prohibited and restricted items in railway passenger security inspection is implemented.

[0035] The fifth aspect of the present application provides a computer program product, including a computer program, which, when executed by a processor, implements the method for intelligent identification of prohibited and restricted items in railway passenger security inspection.

[0036] The present application provides a method for intelligent identification of prohibited and restricted items for railway passenger security inspection, which inputs a target item image corresponding to a target item currently collected and located on a railway passenger security inspection device into a preset prohibited and restricted item intelligent identification model, so that the prohibited and restricted item intelligent identification model outputs prohibited and restricted item identification result data corresponding to the target item image, wherein the prohibited and restricted item intelligent identification model is obtained by pre-training a YOLOv8 optimization model based on various item images of multiple scales and labels corresponding to each of the item images, and the label is used to identify a detection frame on the item image and a prohibited and restricted item identification type identifier to which the detection frame belongs; the YOLOv8 optimization model is pre-formed by optimizing the YOLOv8 model based on the SegNeXt convolution attention mechanism and the GSConv grouping and spatial combined convolution module; outputting the prohibited and restricted item identification result data can effectively reduce the computational complexity of the prohibited and restricted item intelligent identification model and improve the efficiency of the prohibited and restricted item intelligent identification on the basis of ensuring the accuracy of the prohibited and restricted item intelligent identification model, and can effectively improve the accuracy, timeliness and reliability of the prohibited and restricted item intelligent identification, thereby achieving cost reduction and efficiency improvement of the prohibited and restricted item intelligent identification.

[0037] Additional advantages, purposes, and features of the present application will be partially described in the following description, and will become partially apparent to those skilled in the art after studying the following, or may be learned from the practice of the present application. The purposes and other advantages of the present application can be achieved and obtained by the structures specifically pointed out in the specification and the drawings.

[0038] Those skilled in the art will understand that the purposes and advantages that can be achieved by the present application are not limited to the above specific description, and the above and other purposes that can be achieved by the present application will be more clearly understood based on the following detailed description. BRIEF DESCRIPTION OF THE DRAWINGS

[0039] The drawings described herein are used to provide a further understanding of the present application, constitute a part of the present application, and do not constitute a limitation of the present application. The components in the drawings are not drawn to scale, but are only for the purpose of illustrating the principles of the present application. In order to facilitate the illustration and description of some parts of the present application, the corresponding parts in the drawings may be enlarged, that is, they may become larger relative to other components in the exemplary device actually manufactured according to the present application. In the drawings:

[0040] Figure 1 This is a schematic diagram of the first flow chart of the method for intelligent identification of prohibited and restricted items in railway passenger security inspection in one embodiment of the present application.

[0041] Figure 2 This is a second flow chart of the method for intelligent identification of prohibited and restricted items in railway passenger security inspection in one embodiment of the present application.

[0042] Figure 3 Schematic diagram of the architecture of the GSConv module in an example of the present application.

[0043] Figure 4 Schematic diagram of the architecture of the GS bottleneck module in an example of the present application.

[0044] Figure 5 FIG. 1 is a schematic diagram of the architecture of a VoV-GSCSP module in an example of the present application.

[0045] Figure 6 This is a third flow chart of the method for intelligent identification of prohibited and restricted items for railway passenger security inspection in one embodiment of the present application.

[0046] Figure 7 This is a fourth flow chart of the method for intelligent identification of prohibited and restricted items in railway passenger security inspection in one embodiment of the present application.

[0047] Figure 8 This is a schematic diagram of the training process of the intelligent identification model for prohibited and restricted goods in another example of this application.

[0048] Fig. 9 This is a schematic diagram of the structure of a device for intelligently identifying prohibited and restricted items for railway passenger security inspection in one embodiment of the present application. DETAILED DESCRIPTION

[0049] In order to make the purpose, technical solution and advantages of the present application more clearly understood, the present application is further described in detail below in conjunction with the implementation modes and the accompanying drawings. Here, the illustrative implementation modes and descriptions of the present application are used to explain the present application, but are not intended to limit the present application.

[0050] It should also be noted here that in order to avoid obscuring the present application due to unnecessary details, only the structures and / or processing steps closely related to the scheme according to the present application are shown in the accompanying drawings, while other details that are not very relevant to the present application are omitted.

[0051] It should be emphasized that the term “include / comprises” when used herein refers to the presence of features, elements, steps or components, but does not exclude the presence or addition of one or more other features, elements, steps or components.

[0052] It should also be noted that, unless otherwise specified, the term “connection” herein may refer not only to a direct connection but also to an indirect connection involving an intermediate.

[0053] Hereinafter, embodiments of the present application will be described with reference to the accompanying drawings. In the accompanying drawings, the same reference numerals represent the same or similar components, or the same or similar steps.

[0054] In order to solve the problems existing in existing target detection methods, such as being unable to reduce the complexity of the calculation process while meeting the accuracy of the detection results, the embodiments of the present application respectively provide a method for intelligent identification of prohibited and restricted items for railway passenger security inspection, a device for intelligent identification of prohibited and restricted items for railway passenger security inspection, a physical device, a computer-readable storage medium and a computer program product for executing the method for intelligent identification of prohibited and restricted items for railway passenger security inspection, which can effectively reduce the calculation complexity of the intelligent identification model of prohibited and restricted items, and can improve the accuracy and efficiency of intelligent identification of prohibited and restricted items using the model.

[0055] The details are described in detail through the following examples.

[0056] Based on this, the embodiment of the present application provides a method for intelligently identifying prohibited and restricted items for railway passenger security inspection, which can be implemented by an intelligent identification device for prohibited and restricted items for railway passenger security inspection, see Figure 1 The method for intelligent identification of prohibited and restricted items for railway passenger security inspection specifically includes the following contents:

[0057] Step 100: Input the target item image corresponding to the target item currently collected and located on the railway passenger security inspection equipment into a preset prohibited and restricted goods intelligent recognition model, so that the prohibited and restricted goods intelligent recognition model outputs prohibited and restricted goods recognition result data corresponding to the target item image, wherein the prohibited and restricted goods intelligent recognition model is obtained by pre-training a YOLOv8 optimization model based on various item images of multiple scales and labels corresponding to each of the item images, and the label is used to identify a detection frame on the item image and a prohibited and restricted goods recognition type identifier to which the detection frame belongs; the YOLOv8 optimization model is formed by pre-optimizing the YOLOv8 model based on the SegNeXt convolutional attention mechanism and the GSConv grouping and spatial combined convolution module.

[0058] Railway passenger security inspection equipment may be a security inspection device installed in stations and other places, and the security inspection device is provided with an object image shooting device and an object conveyor belt.

[0059] In one or more embodiments of the present application, the label is used to identify the detection frame on the object image and the prohibited or restricted goods identification type identifier to which the detection frame belongs. It can be understood that since not all object images contain prohibited or restricted goods images, the prohibited or restricted goods identification type identifier needs to include at least: an identifier indicating that the object in the detection frame is not a prohibited or restricted item, an identifier that cannot identify whether the object in the detection frame is a prohibited or restricted item, and identifiers respectively used to indicate that the object in the detection frame belongs to different types of prohibited or restricted goods.

[0060] The types of prohibited and restricted items can be set according to specific regulations and actual application requirements. In one example, the types of prohibited and restricted items may include firearms, ammunition and related items, explosives, sharp objects such as controlled knives, flammable and explosive items, poisonous substances, corrosive substances, radioactive substances, infectious pathogens, alcohol and alcoholic beverages, self-spraying pressure vessels, nail polish, power banks and lithium batteries, etc.

[0061] It can be understood that YOLOv8 is a YOLO object detection and image segmentation model. In the YOLOv8 model, the backbone network is responsible for extracting image features, and the detection head is responsible for predicting the location and category of the target. The network structure of the YOLOv8 model is mainly composed of the following three parts connected in sequence:

[0062] 1) Backbone: It uses a series of convolution and deconvolution layers to extract features, and also uses residual connections and bottleneck structures to reduce the size of the network and improve performance. This part uses the C2f module as the basic building block. Compared with the C3 module of YOLOv5, the C2f module has fewer parameters and better feature extraction capabilities.

[0063] 2) Feature Fusion Module (Neck): It uses multi-scale feature fusion technology to fuse feature maps from different stages of Backbone to enhance feature representation capabilities. Specifically, the Neck part of YOLOv8 includes an SPPF module, a PAA module, and two PAN modules.

[0064] 3) Head network: It is responsible for the final target detection and classification tasks, including a detection head and a classification head. The detection head contains a series of convolutional layers and deconvolutional layers to generate detection results; the classification head uses global average pooling to classify each feature map.

[0065] Based on the general architecture of the above-mentioned YOLOv8 model, the YOLOv8 optimization model adopted in the embodiment of the present application is formed by optimizing the YOLOv8 model based on the SegNeXt convolutional attention mechanism and the GSConv grouping and spatial combined convolution module.

[0066] Among them, the SegNeXt convolutional attention mechanism is a simple convolutional neural network structure designed for semantic segmentation tasks. In the field of semantic segmentation, the model needs to effectively encode the contextual information of the image and accurately segment objects of different scales. Although the traditional self-attention mechanism (such as the attention mechanism in Transformer) can capture global information, it is computationally expensive. Therefore, the SegNeXt convolutional attention mechanism proposes a new convolutional attention mechanism that aims to achieve better performance at a lower computational cost.

[0067] Specifically, the SegNeXt attention mechanism can be added to the backbone network (Backbone) of the YOLOv8 model. After inserting the pooling layer SPPF in the backbone network (Backbone) of the YOLOv8 model, the SegNeXt attention mechanism can learn pixel-level attention weights, so that the YOLOv8 optimization model can pay more accurate attention to the area of ​​interest, while ignoring the background area, thereby improving the detection accuracy of the YOLOv8 optimization model.

[0068] Among them, the GSConv (Generalized Shuffle Convolution) grouping and spatial combination convolution module can be referred to as GSConv or GSConv module. GSConv is a lightweight convolution technology that aims to reduce computational costs while maintaining model accuracy. It combines the advantages of standard convolution SC and depth-separable convolution DSC, and achieves channel information fusion through uniform mixing strategy shuffle operation, thereby overcoming the disadvantages of DSC information separation.

[0069] Specifically, the feature fusion module (Neck) of the YOLOv8 model can be optimized by combining GSConv grouping and spatial convolution modules. The Slim-Neck method is introduced through GSConv to reduce the complexity of the original YOLOv8 model while maintaining the accuracy of intelligent identification of prohibited and restricted items in railway passenger security inspection. GSConv can better balance the accuracy and speed of the intelligent identification model of prohibited and restricted items.

[0070] In one or more embodiments of the present application, the prohibited and restricted goods intelligent identification model refers to the YOLOv8 optimization model after model training.

[0071] Step 200: Output the prohibited and restricted goods identification result data.

[0072] In step 200, the prohibited and restricted goods identification result data includes a detection frame on the target item image and a prohibited and restricted goods identification type identifier to which the detection frame belongs. The prohibited and restricted goods identification result data can be directly sent to a display on a railway passenger security inspection instrument for display, and the prohibited and restricted goods identification result data can be sent to a mobile terminal for message prompting. If the prohibited and restricted goods identification type identifier in the prohibited and restricted goods identification result data is displayed as an identifier of a certain prohibited and restricted goods type, then on the basis of outputting the prohibited and restricted goods identification result data, a start signal can also be sent to the early warning device of the security inspection instrument to control the early warning device to send an early warning signal. The early warning device can include at least one of a buzzer, a sound, and a lamp.

[0073] From the above description, it can be seen that the intelligent identification method for prohibited and restricted items in railway passenger security inspection provided in the embodiment of the present application can effectively reduce the computational complexity of the intelligent identification model for prohibited and restricted items and improve the efficiency of intelligent identification of prohibited and restricted items on the basis of ensuring the accuracy of the intelligent identification model for prohibited and restricted items. It can effectively improve the accuracy, timeliness and reliability of intelligent identification of prohibited and restricted items, and achieve cost reduction and efficiency improvement of intelligent identification of prohibited and restricted items.

[0074] In order to further reduce the computational complexity of the prohibited and restricted goods intelligent identification model and improve the efficiency of the prohibited and restricted goods intelligent identification, in a railway passenger security inspection prohibited and restricted goods intelligent identification method provided in an embodiment of the present application, see Figure 2 Step 100 of the method for intelligently identifying prohibited and restricted items for railway passenger security inspection specifically includes the following contents:

[0075] Step 110: Input the target item image corresponding to the target item located on the railway passenger security inspection equipment currently collected into the backbone network based on the SegNeXt convolutional attention mechanism in the preset prohibited and restricted goods intelligent recognition model, so that the backbone network extracts the multi-scale features of the target item image, and then the backbone network inputs the multi-scale features into the feature fusion module provided with GSConv grouping and spatial combined convolution modules in the YOLOv8 model for fusion to obtain the fusion features of the target item image, and then the feature fusion module inputs the fusion features into the head network in the YOLOv8 model, so that the head network outputs the prohibited and restricted goods recognition result data corresponding to the target item image; wherein, the prohibited and restricted goods recognition result data includes a detection frame on the target item image and a prohibited and restricted goods recognition type identifier used to indicate the detection frame.

[0076] Specifically, the SegNeXt attention mechanism mainly includes the following core ideas:

[0077] 1) Multi-Scale Convolutional Attention (MSCA): The MSCA module is a key component of the SegNeXt attention mechanism. It combines depthwise separable convolutions, multi-branch depth-wise strip convolutions, and 1x1 convolutions to capture multi-scale contextual information and inter-channel dependencies.

[0078] 2) Depthwise separable convolution: used to aggregate local information.

[0079] 3) Multi-branch deep strip convolution: used to capture multi-scale contextual information. Each branch uses convolution kernels of different sizes (such as 7x1, 1x7, 11x1, 1x11, etc.) to simulate the effect of large convolution kernels while reducing the amount of computation. 1x1 convolution: used to model the relationship between different channels and output attention weights to reweight the input of MSCA. Lightweight and efficient attention mechanism: Unlike the Transformer-based self-attention mechanism, the SegNeXt attention mechanism uses cheaper and more efficient convolution operations, making the model more computationally efficient.

[0080] 4) Multi-scale feature fusion: SegNeXt not only captures multi-scale features in the encoder, but also fuses these features through an effective attention mechanism to improve the model's ability to segment objects of different scales.

[0081] The specific implementation of the SegNeXt attention mechanism includes the following steps:

[0082] 1) Feature extraction: First, a convolutional neural network (CNN) is used as an encoder to extract the feature representation of the image.

[0083] 2) Application of attention mechanism: Apply the MSCA module on the output feature map of the encoder to generate attention weights. These weights indicate the importance of each pixel or feature.

[0084] 3) Feature fusion: The attention weights are fused with the original feature map through weighted summation or other fusion methods to obtain an enhanced feature representation.

[0085] 4) Decoding and segmentation: Finally, the decoder is used to map the enhanced feature map to pixel-level segmentation results.

[0086] The SegNeXt attention mechanism is added to the Backbone of YOLOv8. After being inserted into the pooling layer SPPF in the yolov8 structure, the attention mechanism learns pixel-level attention weights, allowing the model to pay more attention to the area of ​​interest more accurately and ignore the background area, thereby improving the detection accuracy of the model.

[0087] In addition, since the identification of prohibited and restricted items for railway passenger security inspection is highly time-sensitive, the speed and accuracy of the prohibited and restricted items intelligent identification model are equally important. This application introduces the Slim-Neck method through GSConv to reduce the complexity of the model while maintaining accuracy. GSConv better balances the accuracy and speed of the model.

[0088] See also Figure 3 , the main technical principles of the GSConv module are as follows:

[0089] 1) Combining SC and DSC:

[0090] Standard convolution SC: performs channel-intensive convolution operations to preserve information connections between channels.

[0091] Deep Separable Convolution (DSC): Although the computational cost is low, it has the disadvantage of information separation, that is, the channel information of the input image is separated during calculation, resulting in weak feature extraction and fusion capabilities.

[0092] GSConv first uses SC to perform channel-intensive convolution operations to preserve the information connection between channels, and then combines the advantages of DSC to achieve the goal of reducing the computational cost while maintaining accuracy.

[0093] 2) Shuffle operation:

[0094] The shuffle operation is a uniform mixing strategy that mixes the output of the SC with the output of the DSC to achieve uniform fusion of channel information. Through the shuffle operation, the information of the SC can be completely mixed into the output of the DSC, and the local feature information can be evenly exchanged on different channels.

[0095] 3) Calculate the cost:

[0096] Compared with SC, the computational effort of GSCONV is about 50%, which effectively reduces the computational cost of the model. This gives GSConv a significant advantage in building lightweight models.

[0097] Slim-neck is a design paradigm that aims to improve the performance and computational cost-effectiveness of detectors by optimizing the structure of the neck part. In object detectors, the neck is the part that connects the backbone network and the head network, responsible for feature fusion and processing. By introducing the GSConv module and the GSBottleneck module and VoV-GSCSP module formed based on the GSConv module, Slim-neck is able to reduce computational complexity and inference time while maintaining high accuracy.

[0098] Among them, see Figure 4 GS bottleneck module is an enhanced module based on GSConv, which is used to improve the nonlinear expression of features and the reuse of information. Figure 5 ,The VoV-GSCSP module is a cross-stage partial network module designed using a one-shot aggregation method for ,effective information fusion between feature maps at different stages.

[0099] In order to filter out redundant detection frames and retain the detection frame with the highest confidence as the final result, so as to further improve the accuracy of the intelligent recognition result of prohibited and restricted items and reduce the false detection rate, in a railway passenger security inspection prohibited and restricted items intelligent recognition method provided in an embodiment of the present application, see Figure 2 Step 200 of the method for intelligently identifying prohibited and restricted items for railway passenger security inspection specifically includes the following contents:

[0100] Step 210: If the prohibited and restricted goods recognition result data corresponding to the target item image output by the prohibited and restricted goods intelligent recognition model contains multiple detection frames, one of the multiple detection frames is selected as the target boundary box based on the non-maximum suppression method; if the prohibited and restricted goods recognition result data corresponding to the target item image output by the prohibited and restricted goods intelligent recognition model contains a detection frame, the detection frame is used as the target boundary box.

[0101] Specifically, non-maximum suppression (NMS): when the model outputs multiple detection boxes, the NMS technique is used to filter out redundant detection boxes and retain the detection box with the highest confidence as the final result. This helps to improve the accuracy of the detection results and reduce the false detection rate.

[0102] Step 220: Adjust the target bounding box based on a bounding box regression technique.

[0103] Specifically, bounding box regression: fine-tuning the detected bounding boxes to more accurately locate the location of prohibited items. This can be achieved by using bounding box regression techniques to further improve the accuracy of detection.

[0104] Step 230: Using the target bounding box and the prohibited or restricted item identification type identifier to which the target bounding box belongs as target prohibited or restricted item identification result data for the target item image, and outputting the target prohibited or restricted item identification result data.

[0105] In one example, during the model training stage, based on the multi-scale object images, multi-scale and different-color object images can be further used as data sets for model training, and the generalization ability of the improved model can be tested on data sets of different color styles so that it can adapt to X-ray security inspection devices of various imaging styles.

[0106] In order to further realize non-contact estimation of the length of prohibited and restricted items in an object image on the basis of improving the accuracy of the results of intelligent identification of prohibited and restricted items and reducing the false detection rate, in a method for intelligent identification of prohibited and restricted items for railway passenger security inspection provided in an embodiment of the present application, see Figure 6 The method for intelligently identifying prohibited and restricted items for railway passenger security inspection further specifically includes the following contents before step 100:

[0107] Step 010: Collect X-ray images of the same target item located on the railway passenger security inspection equipment from a first perspective and a second perspective respectively to obtain a first X-ray image corresponding to the first perspective and a second X-ray image corresponding to the second perspective, and use the first X-ray image and the second X-ray image as target item images corresponding to the target item, respectively, wherein the prohibited and restricted items intelligent recognition model outputs prohibited and restricted items recognition result data corresponding to the first X-ray image and the second X-ray image respectively.

[0108] Correspondingly, the following contents are specifically included between step 220 and step 230 in the method for intelligent identification of prohibited and restricted items for railway passenger security inspection:

[0109] Step 020: If the prohibited and restricted goods identification type identifier to which the target boundary box belongs shows that the items in the target boundary box are prohibited and restricted goods, then based on the prohibited and restricted goods identification result data corresponding to the first X-ray image and the second X-ray image, the adjusted target boundary box is subjected to prohibited and restricted goods length identification to obtain the corresponding prohibited and restricted goods length; wherein the prohibited and restricted goods identification result data also includes: the prohibited and restricted goods length.

[0110] In order to further achieve non-contact estimation of the length of sharp objects in an object image, in a method for intelligently identifying prohibited and restricted items for railway passenger security inspection provided in an embodiment of the present application, if the prohibited and restricted item identification type identifier to which the target boundary box belongs includes: sharp objects; see Figure 7 , then step 020 of the method for intelligently identifying prohibited and restricted items for railway passenger security inspection further specifically includes the following contents:

[0111] Step 021: Perform feature detection, image calibration and edge detection on the sharp object in the target boundary boxes corresponding to the first X-ray image and the second X-ray image respectively to determine the image contour of the sharp object.

[0112] Step 022: Measure the pixel length of the image contour of the sharp object, and convert the pixel length into an actual length based on a preset scale to obtain the prohibited item length corresponding to the sharp object.

[0113] Specifically, security inspection hardware devices must also make corresponding adaptation adjustments to achieve rapid and accurate identification of restricted items for railway passenger security inspections, and improve security inspection efficiency and accuracy. Dual-viewing angle X-ray security inspection machines simultaneously collect image data from both horizontal and vertical perspectives to ensure that restricted items can be effectively identified at any angle, avoiding occlusion and missed inspections. Therefore, in the process of detecting restricted items, it is also necessary to combine images from two perspectives. Taking the detection of sharp objects as an example, when a sharp object appears in the X-ray image from a horizontal or vertical perspective, it is determined that there is a sharp object in the package.

[0114] Calculating the actual length of sharp objects by identifying them in X-ray restricted items images is a complex process, because the length information in the image is usually not linearly related to the actual size of the object. Especially when it comes to X-ray images, it is necessary to combine two X-ray images of the same package at the horizontal and vertical viewing angles for size calibration to eliminate image distortion caused by perspective distortion and occlusion factors. The specific method is as follows:

[0115] 1) Feature detection: Extract the feature points of horizontal and vertical X-ray images, such as edges, vertices, intersections, etc., and mark them as feature point set matrices.

[0116] 2) Feature matching: Establish the correspondence between the feature matrices of the two images and use feature description, similarity measurement and feature space relationship to determine the accuracy of the match.

[0117] 3) Transformation model estimation: The corresponding relationship established in feature matching is used to estimate the transformation model, and the linear transformation matrix is ​​obtained using the landmark based method.

[0118] 4) Image resampling and transformation: Use interpolation techniques to achieve image calibration transformation.

[0119] After completing the image calibration, edge detection is used to identify the outline of the knife in the image. Then, the length of the knife outline is accurately measured in pixels on the image, and the pixel length is converted to the actual length using a scale. The scale is calculated as follows:

[0120] 1) Measure the length L1 of the entire package by installing a distance meter at the entrance of the X-ray machine;

[0121] 2) Measure the pixel value L2 corresponding to the body of the bag on the X-ray image;

[0122] 3) The ratio of L1 to L2 is the scale of converting pixel length to actual length.

[0123] In order to further reduce the computational complexity of the prohibited and restricted goods intelligent identification model and improve the efficiency of the prohibited and restricted goods intelligent identification, in a railway passenger security inspection prohibited and restricted goods intelligent identification method provided in an embodiment of the present application, see Figure 6 The method for intelligently identifying prohibited and restricted items for railway passenger security inspection further specifically includes the following contents before step 100:

[0124] Step 030: Obtain a data set consisting of images of various objects at multiple scales.

[0125] And, step 040: optimize the backbone network of the YOLOv8 model based on SegNeXt convolutional attention, and use GSConv grouping and spatial combined convolution modules to optimize the feature fusion module of the YOLOv8 model to obtain the YOLOv8 optimized model.

[0126] Specifically, the specific structure of the backbone network of the YOLOv8 model optimized based on SegNeXt convolutional attention is to add a SegNeXt attention mechanism layer to the last layer of the original YOLOv8 backbone network. Its backbone network has a total of 11 layers, including five convolutional layers connected in sequence, four C2f connection layers, one pooling layer and one attention mechanism layer.

[0127] Specifically, the specific structure of the feature fusion module (Neck) of the YOLOv8 model optimized by the GSConv (Grouped Spatial Convolution) grouping and spatial convolution module includes: a feature extraction layer, a grouped convolution layer, a spatial convolution layer, a channel compression layer, an efficient connection layer and an output layer connected in sequence. Among them, the feature extraction layer generates a preliminary feature map, and then enters the grouped convolution layer to perform grouped convolution on the feature map to reduce the amount of calculation for each group of convolutions while maintaining the ability to extract local features; then the spatial convolution operation is performed, and through cross-group information fusion and information integration, while reducing the number of channels, the integrity and expression of the feature information are maintained as much as possible by optimizing the convolution kernel and the connection method; skip connections and residual connections are used to prevent gradient disappearance and enhance the learning ability of the network; finally, the feature map after the above processing enters the final classification or regression layer.

[0128] Step 050: Use the data set and a preset loss function to train the YOLOv8 optimization model to train the YOLOv8 optimization model into a prohibited and restricted goods intelligent recognition model for outputting prohibited and restricted goods recognition result data corresponding to the object image.

[0129] In order to further improve the effectiveness and reliability of the training data of the prohibited and restricted goods intelligent recognition model, in a railway passenger security inspection prohibited and restricted goods intelligent recognition method provided in an embodiment of the present application, see Figure 7 Step 030 of the method for intelligently identifying prohibited and restricted items for railway passenger security inspection specifically includes the following contents:

[0130] Step 031: Obtain images of various objects at various scales, perform data enhancement and label annotation on each of the object images, so as to obtain a data set consisting of images of various objects at various scales and labels corresponding to each of the object images.

[0131] Specifically, you can first build a dedicated data set: For the task of identifying prohibited and restricted items for railway security passenger transportation, you need to build a data set containing images of various prohibited and restricted items. These images should cover different angles, lighting conditions, occlusion conditions, etc. to improve the generalization ability of the model. Then perform data enhancement: Use data enhancement techniques such as rotation, scaling, translation, flipping, and color jitter to increase the diversity of the training set and help the model learn more robust feature representations. Improve the accuracy of annotations: Use high-precision annotation tools to accurately annotate prohibited and restricted items in the image, including the precise location of the bounding box and the category label. Ensuring the accuracy and consistency of annotations is crucial to improving the recognition accuracy of the model.

[0132] In a specific example, the present application also provides a training process of a prohibited and restricted goods intelligent identification model in a railway passenger security inspection prohibited and restricted goods intelligent identification method, see Figure 8 The training process of the prohibited and restricted goods intelligent identification model specifically includes the following contents:

[0133] (1) Data preparation

[0134] 1) Build a dedicated dataset: For the task of identifying prohibited and restricted items for railway security inspection, it is necessary to build a dataset containing images of various prohibited and restricted items. These images should cover different angles, lighting conditions, occlusion conditions, etc., to improve the generalization ability of the model.

[0135] 2) Data enhancement: Use data enhancement techniques such as rotation, scaling, translation, flipping, and color jittering to increase the diversity of the training set and help the model learn more robust feature representations.

[0136] 3) Annotation accuracy: Use high-precision annotation tools to accurately annotate prohibited and restricted items in images, including the precise location of the bounding box and the category label. Ensuring the accuracy and consistency of annotation is critical to improving the recognition accuracy of the model.

[0137] (2) Model Optimization

[0138] 1) Backbone network selection: YOLOv8 uses CSPDarknet53 as the backbone network by default, but you can consider replacing it with a more powerful backbone network, such as ResNet or EfficientNet, to improve the feature extraction capability of the model.

[0139] 2) Feature fusion: Optimize feature fusion modules such as PANet and FPN to ensure that feature information of different scales is effectively integrated, thereby improving the model's ability to detect prohibited and restricted items of different sizes.

[0140] 3) Loss function optimization: YOLOv8 uses the CIoU loss function to optimize the regression of the bounding box. You can consider adjusting the weight of the CIoU loss function, or try other loss functions that are more suitable for object detection tasks, such as Focal Loss, GIoU Loss, etc., to further improve the precision and recall of the model.

[0141] 4) Prior box optimization: According to the size distribution of prohibited and restricted items in the data set, the setting of the prior box is optimized to make it more suitable for the needs of the object detection task. This helps the model to more accurately predict the location and size of the bounding box.

[0142] (3) Training strategy

[0143] 1) Learning rate and optimizer adjustment: Use a suitable learning rate warm-up strategy and dynamically adjust the learning rate according to the convergence of the model during training. Select a suitable optimizer (such as Adam) and adjust its parameters to speed up the model training and improve the convergence performance.

[0144] 2) Multi-scale training: Use images of different sizes for training to enhance the model’s adaptability to objects of different scales. This can be achieved by randomly changing the size of the input image during training.

[0145] 3) Regularization and dropout: Regularization terms (such as L1 and L2 regularization) and dropout mechanisms are introduced into the model to reduce overfitting and improve the generalization ability of the model.

[0146] (4) Post-processing technology

[0147] 1) Non-maximum suppression (NMS): When the model outputs multiple detection boxes, the NMS technique is used to filter out redundant detection boxes and retain the detection box with the highest confidence as the final result. This helps to improve the accuracy of the detection results and reduce the false detection rate.

[0148] 2) Bounding Box Regression: Fine-tune the detected bounding boxes to more accurately locate the location of prohibited and restricted items. This can be achieved by using bounding box regression techniques to further improve the accuracy of detection.

[0149] Finally, the generalization ability of the improved model is tested on data sets with different color styles, so that it can adapt to X-ray security inspection devices with various imaging styles.

[0150] From the software level, the present application also provides a railway passenger security inspection prohibited and restricted goods intelligent identification device for executing all or part of the railway passenger security inspection prohibited and restricted goods intelligent identification method, see Fig. 9 The railway passenger security inspection prohibited and restricted items intelligent identification device specifically includes the following contents:

[0151] The model recognition module 10 is used to input the target item image corresponding to the target item currently collected and located on the railway passenger security inspection equipment into a preset prohibited and restricted goods intelligent recognition model, so that the prohibited and restricted goods intelligent recognition model outputs prohibited and restricted goods recognition result data corresponding to the target item image, wherein the prohibited and restricted goods intelligent recognition model is obtained by training a YOLOv8 optimization model based on various item images of multiple scales and labels corresponding to each of the item images, and the label is used to identify the detection box on the item image and the prohibited and restricted goods recognition type identifier to which the detection box belongs; the YOLOv8 optimization model is formed by pre-optimizing the YOLOv8 model based on the SegNeXt convolutional attention mechanism and the GSConv grouping and spatial combined convolution module.

[0152] The result output module 20 is used to output the prohibited and restricted goods identification result data.

[0153] The embodiment of the device for intelligent identification of prohibited and restricted items for railway passenger security inspection provided in the present application can be specifically used to execute the processing flow of the embodiment of the method for intelligent identification of prohibited and restricted items for railway passenger security inspection in the above-mentioned embodiment. Its functions will not be repeated here, and reference can be made to the detailed description of the embodiment of the method for intelligent identification of prohibited and restricted items for railway passenger security inspection mentioned above.

[0154] The part of the intelligent identification of prohibited and restricted items for railway passenger security inspection by the intelligent identification device for railway passenger security inspection can be executed in the server or completed in the client device. The specific selection can be based on the processing capacity of the client device and the limitations of the user's usage scenario. This application is not limited to this. If all operations are completed in the client device, the client device may also include a processor for specific processing of intelligent identification of prohibited and restricted items for railway passenger security inspection.

[0155] The client device may have a communication module (i.e., a communication unit) that can communicate with a remote server to achieve data transmission with the server. The server may include a server on the task scheduling center side, and other implementation scenarios may also include a server on an intermediate platform, such as a server on a third-party server platform that has a communication link with the task scheduling center server. The server may include a single computer device, or a server cluster consisting of multiple servers, or a server structure of a distributed device.

[0156] The server and the client device may communicate with each other using any suitable network protocol, including network protocols that have not yet been developed on the date of filing this application. The network protocols may include, for example, TCP / IP, UDP / IP, HTTP, HTTPS, etc. Of course, the network protocols may also include, for example, RPC (Remote Procedure Call Protocol) and REST (Representational State Transfer) protocols used on top of the above protocols.

[0157] From the above description, it can be seen that the railway passenger security prohibited and restricted goods intelligent identification device provided in the embodiment of the present application can effectively reduce the computational complexity of the prohibited and restricted goods intelligent identification model and improve the efficiency of the prohibited and restricted goods intelligent identification on the basis of ensuring the accuracy of the prohibited and restricted goods intelligent identification model. It can effectively improve the accuracy, timeliness and reliability of the prohibited and restricted goods intelligent identification, and achieve cost reduction and efficiency improvement of the prohibited and restricted goods intelligent identification.

[0158] The embodiment of the present application also provides an electronic device, which may include a processor, a memory, a receiver and a transmitter, wherein the processor is used to execute the railway passenger security inspection prohibited and restricted goods intelligent identification method mentioned in the above embodiment, wherein the processor and the memory may be connected via a bus or other means, taking the bus connection as an example. The receiver may be connected to the processor and the memory via a wired or wireless manner.

[0159] The processor may be a central processing unit (CPU). The processor may also be other general-purpose processors, digital signal processors (DSP), application-specific integrated circuits (ASIC), field-programmable gate arrays (FPGA) or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, or a combination of the above chips.

[0160] As a non-transient computer-readable storage medium, the memory can be used to store non-transient software programs, non-transient computer executable programs and modules, such as the program instructions / modules corresponding to the method for intelligent identification of prohibited and restricted items for railway passenger security inspection in the embodiment of the present application. The processor executes various functional applications and data processing of the processor by running the non-transient software programs, instructions and modules stored in the memory, that is, the method for intelligent identification of prohibited and restricted items for railway passenger security inspection in the above method embodiment is implemented.

[0161] The memory may include a program storage area and a data storage area, wherein the program storage area may store an operating system, an application required for at least one function; the data storage area may store data created by the processor, etc. In addition, the memory may include a high-speed random access memory, and may also include a non-volatile memory, such as at least one disk storage device, a flash memory device, or other non-volatile solid-state storage device. In some embodiments, the memory may optionally include a memory remotely arranged relative to the processor, and these remote memories may be connected to the processor via a network. Examples of the above-mentioned network include, but are not limited to, the Internet, an intranet, a local area network, a mobile communication network, and combinations thereof.

[0162] The one or more modules are stored in the memory, and when executed by the processor, the method for intelligent identification of prohibited and restricted items for railway passenger security inspection in the embodiment is executed.

[0163] In some embodiments of the present application, the user equipment may include a processor, a memory, and a transceiver unit, which may include a receiver and a transmitter. The processor, memory, receiver, and transmitter may be connected through a bus system. The memory is used to store computer instructions, and the processor is used to execute the computer instructions stored in the memory to control the transceiver unit to send and receive signals.

[0164] As an implementation method, the functions of the receiver and the transmitter in the present application can be considered to be implemented through a transceiver circuit or a dedicated chip for transceiver, and the processor can be considered to be implemented through a dedicated processing chip, a processing circuit or a general chip.

[0165] As another implementation method, it is possible to use a general-purpose computer to implement the server provided in the embodiment of the present application, that is, to store the program code for implementing the functions of the processor, receiver, and transmitter in a memory, and the general-purpose processor implements the functions of the processor, receiver, and transmitter by executing the code in the memory.

[0166] The embodiment of the present application also provides a computer-readable storage medium, on which a computer program is stored, and when the computer program is executed by a processor, the steps of the aforementioned railway passenger security inspection prohibited and restricted items intelligent identification method are implemented. The computer-readable storage medium can be a tangible storage medium, such as a random access memory (RAM), a memory, a read-only memory (ROM), an electrically programmable ROM, an electrically erasable programmable ROM, a register, a floppy disk, a hard disk, a removable storage disk, a CD-ROM, or any other form of storage medium known in the technical field.

[0167] An embodiment of the present application also provides a computer program product, which, when executed by a processor, implements the steps of the aforementioned method for intelligent identification of prohibited and restricted items in railway passenger security inspection.

[0168] It should be understood by those skilled in the art that the exemplary components, systems and methods described in conjunction with the embodiments disclosed herein can be implemented in hardware, software or a combination of the two. Whether it is specifically performed in hardware or software depends on the specific application and design constraints of the technical solution. Professional and technical personnel can use different methods to implement the described functions for each specific application, but such implementation should not be considered to be beyond the scope of this application. When implemented in hardware, it can be, for example, an electronic circuit, an application specific integrated circuit (ASIC), appropriate firmware, a plug-in, a function card, etc. When implemented in software, the elements of this application are programs or code segments used to perform the required tasks. The program or code segment can be stored in a machine-readable medium, or transmitted on a transmission medium or a communication link via a data signal carried in a carrier.

[0169] It should be clear that the present application is not limited to the specific configuration and processing described above and shown in the figures. For the sake of simplicity, a detailed description of the known method is omitted here. In the above embodiments, several specific steps are described and shown as examples. However, the method process of the present application is not limited to the specific steps described and shown, and those skilled in the art can make various changes, modifications and additions, or change the order between the steps after understanding the spirit of the present application.

[0170] In the present application, features described and / or illustrated for one embodiment may be used in the same manner or in a similar manner in one or more other embodiments, and / or combined with features of other embodiments or replace features of other embodiments.

[0171] The above description is only the preferred embodiment of the present application and is not intended to limit the present application. For those skilled in the art, the embodiments of the present application may have various modifications and variations. Any modification, equivalent replacement, improvement, etc. made within the spirit and principle of the present application shall be included in the protection scope of the present application.

Claims

1. A method for intelligent identification of prohibited and restricted items for railway passenger security inspection, characterized in that: include: Input the target item image corresponding to the target item currently collected and located on the railway passenger security inspection equipment into the preset prohibited and restricted goods intelligent recognition model, so that the prohibited and restricted goods intelligent recognition model outputs the prohibited and restricted goods recognition result data corresponding to the target item image, wherein the prohibited and restricted goods intelligent recognition model is obtained by training the YOLOv8 optimization model based on various item images of multiple scales and the labels corresponding to each of the item images, and the labels are used to identify the detection frame on the item image and the prohibited and restricted goods recognition type identifier to which the detection frame belongs; the YOLOv8 optimization model is formed by optimizing the YOLOv8 model based on the SegNeXt convolution attention mechanism and the GSConv grouping and spatial combined convolution module in advance; Output the prohibited and restricted goods identification result data.

2. The method for intelligent identification of prohibited and restricted items for railway passenger security inspection according to claim 1 is characterized in that: The step of inputting the target item image corresponding to the target item located on the railway passenger security inspection device currently collected into a preset prohibited and restricted goods intelligent recognition model so that the prohibited and restricted goods intelligent recognition model outputs prohibited and restricted goods recognition result data corresponding to the target item image includes: The target item image corresponding to the target item currently collected and located on the railway passenger security inspection equipment is input into the backbone network based on the SegNeXt convolutional attention mechanism in the preset prohibited and restricted goods intelligent recognition model, so that the backbone network extracts the multi-scale features of the target item image, and then the backbone network inputs the multi-scale features into the feature fusion module provided with GSConv grouping and spatial combined convolution modules in the YOLOv8 model for fusion to obtain the fusion features of the target item image, and then the feature fusion module inputs the fusion features into the head network in the YOLOv8 model, so that the head network outputs the prohibited and restricted goods recognition result data corresponding to the target item image; wherein, the prohibited and restricted goods recognition result data includes a detection frame on the target item image and a prohibited and restricted goods recognition type identifier used to indicate the detection frame.

3. The method for intelligent identification of prohibited and restricted items for railway passenger security inspection according to claim 1 is characterized in that: The outputting of the prohibited and restricted goods identification result data includes: If the prohibited and restricted goods recognition result data corresponding to the target item image output by the prohibited and restricted goods intelligent recognition model contains multiple detection frames, one of the multiple detection frames is selected as the target boundary frame based on the non-maximum suppression method; if the prohibited and restricted goods recognition result data corresponding to the target item image output by the prohibited and restricted goods intelligent recognition model contains one detection frame, the detection frame is used as the target boundary frame; Adjusting the target bounding box based on a bounding box regression technique; The target bounding box and the prohibited or restricted item identification type identifier to which the target bounding box belongs are used as target prohibited or restricted item identification result data for the target item image, and the target prohibited or restricted item identification result data is output.

4. The method for intelligent identification of prohibited and restricted items for railway passenger security inspection according to claim 3 is characterized in that: Before inputting the currently collected target item image located on the railway passenger security inspection device into the preset prohibited and restricted item intelligent recognition model, the method further includes: Acquire X-ray images of the same target object located on the railway passenger security inspection device from a first viewing angle and a second viewing angle, respectively, to obtain a first X-ray image corresponding to the first viewing angle and a second X-ray image corresponding to the second viewing angle, and use the first X-ray image and the second X-ray image as target object images corresponding to the target object, respectively; The prohibited and restricted goods intelligent identification model outputs the prohibited and restricted goods identification result data corresponding to the first X-ray image and the second X-ray image respectively; Correspondingly, before outputting the prohibited and restricted goods identification result data, the method further includes: If the prohibited or restricted item identification type identifier to which the target boundary box belongs indicates that the item in the target boundary box is a prohibited or restricted item, then, according to the prohibited or restricted item identification result data corresponding to each of the first X-ray image and the second X-ray image, performing prohibited or restricted item length identification on the adjusted target boundary box to obtain a corresponding prohibited or restricted item length; Among them, the prohibited and restricted goods identification result data also includes: the length of the prohibited and restricted goods.

5. The method for intelligent identification of prohibited and restricted items for railway passenger security inspection according to claim 4 is characterized in that: If the prohibited and restricted goods identification type identifier to which the target boundary box belongs includes: sharp objects; then, according to the prohibited and restricted goods identification result data corresponding to the first X-ray image and the second X-ray image, the prohibited and restricted goods length identification is performed on the adjusted target boundary box to obtain the corresponding prohibited and restricted goods length, including: Performing feature detection, image calibration and edge detection on the sharp object in the target boundary boxes corresponding to the first X-ray image and the second X-ray image, respectively, to determine the image contour of the sharp object; The pixel length of the image contour of the sharp object is measured, and the pixel length is converted into an actual length based on a preset scale to obtain the prohibited item length corresponding to the sharp object.

6. The method for intelligent identification of prohibited and restricted items for railway passenger security inspection according to claim 1 is characterized in that: Before inputting the currently collected target item image located on the railway passenger security inspection device into the preset prohibited and restricted item intelligent recognition model, the method further includes: Obtain a dataset consisting of images of various objects at various scales; In addition, the backbone network of the YOLOv8 model is optimized based on SegNeXt convolutional attention, and the feature fusion module of the YOLOv8 model is optimized using GSConv grouping and spatial combined convolution modules to obtain the YOLOv8 optimized model; The YOLOv8 optimization model is trained using the data set and a preset loss function to train the YOLOv8 optimization model into a prohibited and restricted goods intelligent recognition model for outputting prohibited and restricted goods recognition result data corresponding to the object image.

7. The method for intelligently identifying prohibited and restricted items for railway passenger security inspection according to claim 6 is characterized in that: The step of obtaining a data set consisting of images of various objects at various scales includes: Obtaining images of various objects at various scales, performing data enhancement and labeling on the various object images, so as to obtain a data set consisting of the various object images at various scales and labels corresponding to the various object images.

8. An intelligent identification device for prohibited and restricted items for railway passenger security inspection, characterized in that: include: A model recognition module is used to input the target item image corresponding to the target item currently collected and located on the railway passenger security inspection equipment into a preset prohibited and restricted goods intelligent recognition model, so that the prohibited and restricted goods intelligent recognition model outputs prohibited and restricted goods recognition result data corresponding to the target item image, wherein the prohibited and restricted goods intelligent recognition model is obtained by training a YOLOv8 optimization model based on various item images of multiple scales and labels corresponding to each of the item images, and the label is used to identify a detection frame on the item image and a prohibited and restricted goods recognition type identifier to which the detection frame belongs; the YOLOv8 optimization model is formed by optimizing the YOLOv8 model based on the SegNeXt convolutional attention mechanism and the GSConv grouping and spatial combined convolution module in advance; The result output module is used to output the prohibited and restricted goods identification result data.

9. An electronic device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that: When the processor executes the computer program, the method for intelligently identifying prohibited and restricted items for railway passenger security inspection as described in any one of claims 1 to 7 is implemented.

10. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the method for intelligently identifying prohibited and restricted items for railway passenger security inspection as described in any one of claims 1 to 7 is implemented.

Images