Hazardous Chemical Identification System Based on Deep Learning and Automatic Image Recognition and Classification

Abstract

The present invention provides a hazardous chemical identification system based on deep learning and automatic image recognition and classification, which includes multiple functional modules. During the identification process, multiple perspective images of the target area are obtained. By performing operations such as cropping and enhancement on the multiple perspective images, the target images for classification and recognition are determined. The target images are input into a deep learning model. After being processed by the deep learning model, the object features of each target object contained in the target images are extracted. Based on the object features of each target object, it is identified whether the target object is a hazardous chemical. When it is identified that the target object is a hazardous chemical, the chemical type of the hazardous chemical to which it belongs is output. When using the system provided by the present invention to identify and classify hazardous chemicals, the images taken of the area to be identified are input into the deep learning model. After being processed by the deep learning model, it can be obtained whether there are hazardous chemicals in the relevant area and the relevant classification, improving the efficiency and accuracy of hazardous chemical identification.

Description

Technical Field

[0001] The present invention relates to the field of computer technology, and particularly to a hazardous chemical identification system based on deep learning and automatic image recognition and classification. Background Art

[0002] Hazardous chemicals refer to highly toxic chemicals and other chemicals with properties such as toxicity, corrosion, explosion, combustion, and combustion support, which are harmful to humans, facilities, and the environment. Given the hazards of hazardous chemicals, timely identification of hazardous chemicals is required in public safety fields such as daily life and transportation.

[0003] Traditional hazardous chemical identification methods mostly rely on manual inspection or handheld detection equipment, which have problems such as low efficiency and high misjudgment rate. Summary of the Invention

[0004] In view of this, the present invention provides a hazardous chemical identification system based on deep learning and automatic image recognition and classification, which can quickly identify and classify hazardous chemicals, improve the identification efficiency, and reduce the misjudgment rate.

[0005] A hazardous chemical identification system based on deep learning and automatic image recognition and classification includes:

[0006] A receiving module, configured to receive perspective images obtained by a detection device taking multi-angle shots of a target area;

[0007] A selection module, configured to select, from the perspective images, the perspective image taken from the front view as a reference image;

[0008] A determination module, configured to determine, based on the perspective images, an object area included in the reference image;

[0009] A cropping module, configured to crop the image where the object area is located as an image to be recognized;

[0010] An enhancement module, configured to perform enhancement processing on the image to be recognized to obtain a target image corresponding to the image to be recognized;

[0011] An identification module, configured to input the target image into a pre-established deep learning model, process the target image through the deep learning model, determine each target object included in the target image and the object form of each target object, extract the object features of each target object based on the object form of each target object, and identify whether the target object is a hazardous chemical based on the object features of each target object. When it is recognized that the target object is a hazardous chemical, output the chemical type of the hazardous chemical to which the target object belongs;

[0012] A training module, configured to determine a plurality of test images, where the plurality of test images include images of various types of hazardous chemicals; perform enhancement processing on the plurality of test images to obtain a plurality of sample images; determine each object included in each of the sample images, and add an attribute label to each object in each of the sample images; extract the object features of each object in each of the sample images, where the object features at least include the shape, color, texture, and edges of the object; determine an initial learning model, and use the respective object features of each object in each of the sample images as inputs and the attribute labels of each object in each of the sample images as outputs to train the initial learning model until the convergence function of the initial learning model is less than a preset threshold, at which point the training of the initial learning model is completed to establish the deep learning model; each of the attribute labels is used to characterize whether the object corresponding to it is a hazardous chemical, and when the object is a hazardous chemical, the chemical type of the hazardous chemical to which the object belongs.

[0013] Optionally, the above-mentioned hazardous chemical identification system based on deep learning and automatic image recognition and classification further includes:

[0014] A detection device, configured to capture the target area from multiple angles to obtain perspective images at each angle.

[0015] Optionally, in the above-mentioned hazardous chemical identification system based on deep learning and automatic image recognition and classification, the determination module includes:

[0016] An analysis sub-module, configured to perform object analysis on each of the perspective images to determine the object areas where objects exist in each perspective.

[0017] A mapping sub-module, configured to map the object areas where objects exist in each perspective to the reference image to determine the object areas included in the reference image.

[0018] Optionally, in the above-mentioned hazardous chemical identification system based on deep learning and automatic image recognition and classification, the interception module includes:

[0019] A judgment sub-module, configured to judge whether the object areas included in the reference image are continuous areas.

[0020] An interception sub-module, configured to, if the object areas included in the reference image are not continuous areas, respectively intercept the images corresponding to each discrete object area included in the reference image, and use the intercepted images as the images to be recognized.

[0021] Optionally, in the above-mentioned hazardous chemical identification system based on deep learning and automatic image recognition and classification, the enhancement module includes:

[0022] Enhancement module, used to perform enhancement processing on the image to be recognized, and the enhancement processing includes at least any one or any combination of grayscale conversion, binarization, denoising operation, and contrast enhancement.

[0023] The above-mentioned hazardous chemical identification system based on deep learning and automatic image recognition and classification, optionally, further includes:

[0024] Display module, used to show the user the recognition result of whether the target object output by the deep learning model is a hazardous chemical, and when the target object is a hazardous chemical, the chemical type of the target object as the hazardous chemical.

[0025] The above-mentioned hazardous chemical identification system based on deep learning and automatic image recognition and classification, optionally, the training module includes:

[0026] Cleaning sub-module, used to clean the multiple test images and remove the blurred images and duplicate images in the multiple test images.

[0027] The above-mentioned hazardous chemical identification system based on deep learning and automatic image recognition and classification, optionally, further includes:

[0028] Model testing module, used to test the established deep learning model to determine the processing performance of the deep learning model.

[0029] The above-mentioned hazardous chemical identification system based on deep learning and automatic image recognition and classification, optionally, further includes:

[0030] Model optimization module, used to optimize the established deep learning model to adjust the model structure and hyperparameters of the deep learning model.

[0031] The above-mentioned hazardous chemical identification system based on deep learning and automatic image recognition and classification, optionally, further includes:

[0032] Model deployment interface, used to connect the deep learning model to a cloud server or an edge device to complete the deployment of the deep learning model in the cloud server or the edge device.

[0033] Compared with the prior art, the present invention has the following advantages:

[0034] The present invention provides a hazardous chemical identification system based on deep learning and automatic image recognition and classification, comprising: a receiving module for receiving perspective images obtained by a detection device through multi-angle shooting of a target area; a selection module for selecting, from the perspective images, the perspective image taken from the front view as a reference image; a determination module for determining, based on the perspective images, the object area included in the reference image; a cropping module for cropping the image where the object area is located as an image to be recognized; an enhancement module for performing enhancement processing on the image to be recognized to obtain a target image corresponding to the image to be recognized; a recognition module for inputting the target image into an established deep learning model, processing the target image through the deep learning model, determining each target object included in the target image and the object form of each target object, and extracting the object features of each target object based on the object form of each target object, and identifying whether the target object is a hazardous chemical based on the object features of each target object, and when it is recognized that the target object is a hazardous chemical, outputting the chemical type of the hazardous chemical to which the target object belongs; a training module for determining a plurality of test images, the plurality of test images including images of various types of hazardous chemicals; performing enhancement processing on the plurality of test images to obtain a plurality of sample images; determining each object included in each sample image and adding an attribute label to each object in each sample image; extracting the object features of each object in each sample image, the object features at least including the form, color, texture and edge of the object; determining an initial learning model, training the initial learning model with the object features of each object in each sample image as input and the attribute label of each object in each sample image as output until the convergence function of the initial learning model is less than a preset threshold, completing the training of the initial learning model to establish the deep learning model; each attribute label being used to characterize whether the object corresponding thereto is a hazardous chemical and, when the object is a hazardous chemical, the chemical type of the hazardous chemical to which the object belongs. When the system provided by the present invention is used to identify and classify hazardous chemicals, the images taken of the area to be recognized are input into the deep learning model, and through the processing of the deep learning model, it can be obtained whether there are hazardous chemicals in the relevant area and the relevant classification, improving the efficiency and accuracy of hazardous chemical identification. BRIEF DESCRIPTION OF THE DRAWINGS

[0035] In order to more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the following will briefly introduce the drawings required for the description of the embodiments or the prior art. Obviously, the drawings in the following description are only the embodiments of the present invention, and those of ordinary skill in the art can also obtain other drawings according to the provided drawings without creative efforts.

[0036] Figure 1 The structural schematic diagram of a hazardous chemical identification system based on deep learning and automatic image recognition and classification provided by an embodiment of the present invention;

[0037] Figure 2 Another structural schematic diagram of a hazardous chemical identification system based on deep learning and automatic image recognition and classification provided by an embodiment of the present invention. Detailed implementation manners

[0038] Next, the technical solutions in the embodiments of the present invention will be clearly and completely described in conjunction with the accompanying drawings in the embodiments of the present invention. Obviously, the described embodiments are only a part of the embodiments of the present invention, rather than all the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those of ordinary skill in the art without creative efforts shall fall within the protection scope of the present invention.

[0039] In this application, relational terms such as first and second are only used to distinguish one entity or operation from another entity or operation, and do not necessarily require or imply any actual relationship or order between these entities or operations. The term "comprising", "including" or any other variant thereof is intended to cover non-exclusive inclusion, so that a process, method, article or device including a series of elements not only includes those elements, but also includes other elements not expressly listed, or also includes elements inherent to such process, method, article or device. Without further limitation, an element defined by the statement "including one..." does not exclude the presence of additional identical elements in the process, method, article or device including the element.

[0040] The present invention can be used in many general or special computing device environments or configurations. For example: personal computers, server computers, handheld devices or portable devices, tablet devices, multi-processor devices, distributed computing environments including any of the above devices or equipment, and so on.

[0041] An embodiment of the present invention provides a hazardous chemical identification system based on deep learning and automatic image recognition and classification. The hazardous chemical identification system can be applied in fields such as public safety, industrial production, and transportation hubs, and can be set in the cloud servers corresponding to the above-mentioned various fields. The structural schematic diagram of the hazardous chemical identification system based on deep learning and automatic image recognition and classification provided by an embodiment of the present invention is as Figure 1 shown and includes:

[0042] A receiving module 101, configured to receive each perspective image obtained by a detection device taking multi-angle shots of a target area;

[0043] In an embodiment of the present invention, the receiving module 101 receives in real time each perspective image obtained by a detection device through multi-angle photographing of a target area.

[0044] The detection device can be various different types of camera devices. In an embodiment of the present invention, images in different formats collected by different types of camera devices can be processed.

[0045] In an embodiment of the present invention, the target area can be a fixed area, such as the security inspection machine area in the field of public transportation, or a necessary area that a vehicle needs to pass through in the field of transportation. The detection device is arranged at various set positions in the target area. When items such as the user's luggage, goods transported by the vehicle, or individuals entering or leaving important places pass through the target area, the detection device performs multi-angle photographing on the items or individuals passing through the target area to obtain perspective images at each angle.

[0046] The receiving module 101 receives each perspective image obtained by the detection device and transmits each perspective image to the selection module 102.

[0047] The selection module 102 is used to select, from each of the perspective images, the perspective image taken from the front view angle as the reference image.

[0048] In an embodiment of the present invention, the selection module 102 performs perspective analysis on each perspective image received by the receiving module 101, determines the front view angle according to the scene where the current target area is located. In the actual operation process, such as in the security inspection area of a station or an airport, the front shooting angle from top to bottom of the detection device can be determined as the front view angle. During the transportation of large goods, the angle parallel to the large goods can also be set as the official angle. In an embodiment of the present invention, the front view angle is determined according to the application scenario where the target area is located.

[0049] The selection module 102 selects, according to the determined front view angle, the perspective image that matches the determined front view angle from each of the perspective images as the reference image.

[0050] The determination module 103 is used to determine the object area included in the reference image based on each of the perspective images.

[0051] In an embodiment of the present invention, during the actual photographing process of the detection device, since the placement of each object in the luggage or goods is irregular and there is an occlusion situation, when determining the object area in the reference image, the object area included in the reference image can be comprehensively determined based on the perspective images from different perspectives.

[0052] In the actual application process, the embodiments of the present invention can identify dangerous chemicals in multiple scenarios. For example, during the infrared ray security inspection process, due to the characteristics of infrared rays, the object areas where objects exist as a whole can be identified from multiple angles.

[0053] During the image capture process, such as when capturing images of various objects exposed in the actual environment, there may be occluded areas. In this case, it is necessary to comprehensively consider the perspective images from various angles to determine the object regions in the reference image.

[0054] The cropping module 104 is configured to crop the image where the object region is located as the image to be recognized.

[0055] In an embodiment of the present invention, the cropping module 104 crops the image of the object region where there is an object in the reference image, removes the part of the reference image that does not contain an object, and uses the image of the part where there is an object as the image to be recognized.

[0056] The enhancement module 105 is configured to perform enhancement processing on the image to be recognized to obtain the target image corresponding to the image to be recognized.

[0057] In an embodiment of the present invention, the enhancement module 105 performs enhancement processing on the image to be recognized to improve the processability of the image to be recognized and obtain the target image corresponding to the image to be recognized. This target image is the image input into the established deep learning model for recognition processing.

[0058] The recognition module 106 is configured to input the target image into the established deep learning model. After being processed by the deep learning model, it determines each target object included in the target image and the object form of each target object, and extracts the object features of each target object based on the object form of each target object. It then identifies whether the target object is a hazardous chemical based on the object features of each target object. When it is recognized that the target object is a hazardous chemical, it outputs the chemical type of the hazardous chemical to which the target object belongs.

[0059] In an embodiment of the present invention, the recognition module 106 inputs the target image into the established deep learning model. In this deep learning model, it further determines each target object included in the target image. Through the object characteristics of different objects, the various target objects in the target image can be distinguished. At the same time, the deep learning model extracts the object features of each target object, and based on the features of each target object, it can identify whether the target object is a hazardous chemical. Further, when some target objects are recognized as hazardous chemicals, it further determines the type of the hazardous chemical to which the target object belongs.

[0060] The training module 107 is used to determine a plurality of test images, where the plurality of test images include images of various types of hazardous chemicals; perform enhancement processing on the plurality of test images to obtain a plurality of sample images; determine each object included in each of the sample images, and add an attribute label to each object in each of the sample images; extract the object features of each object in each of the sample images, where the object features at least include the shape, color, texture, and edges of the object; determine an initial learning model, use the respective object features of each object in each of the sample images as inputs, and use the attribute labels of each object in each of the sample images as outputs to train the initial learning model until the convergence function of the initial learning model is less than a preset threshold, at which point the training of the initial learning model is completed to establish the deep learning model; each of the attribute labels is used to characterize whether the object corresponding to it is a hazardous chemical, and when the object is a hazardous chemical, the chemical type of the hazardous chemical to which the object belongs.

[0061] In an embodiment of the present invention, during the process of obtaining a deep learning model through training, first, a plurality of test images are determined. Each test image may include an image of a type of hazardous chemical. The acquisition process of the test images can be obtained by using test devices such as cameras or infrared acquisition devices to collect images of hazardous chemicals existing in different scenarios.

[0062] In an embodiment of the present invention, the plurality of test images may include all categories of various hazardous chemicals. After performing enhancement processing on the plurality of test images, each sample image used for training the deep learning model is determined.

[0063] An attribute label is added to each object in each sample image. If the object in the sample image does not belong to a hazardous chemical, the attribute label may label the name and classification of the object and that the object does not belong to a hazardous chemical.

[0064] If there is a hazardous chemical among the objects included in the sample image, the attribute label for the object may label that the object belongs to a hazardous chemical and label the classification and hazard level of the hazardous chemical.

[0065] In an embodiment of the present invention, using the respective object features of each object in each sample image as inputs and the attribute labels of the objects in each sample image as outputs, the determined initial model is trained until the convergence value of the convergence function of the initial model is less than a preset threshold, at which point the training of the initial model ends and a deep learning model is obtained.

[0066] The hazardous chemical identification system based on deep learning and automatic image recognition and classification provided by the embodiments of the present invention, when applying the system provided by the present invention to identify and classify hazardous chemicals, inputs the images taken in the area to be identified into the deep learning model. After being processed by the deep learning model, it can obtain whether there are hazardous chemicals in the relevant area and the relevant classification, improving the efficiency and accuracy of hazardous chemical identification.

[0067] Reference Figure 2 Fig. shows another detailed structural schematic diagram of the hazardous chemical identification system based on deep learning and automatic image recognition and classification provided by the embodiments of the present invention. In Figure 2 the system provided by the embodiments of the present invention further includes:

[0068] The detection device 108 is used to take pictures of the target area from multiple angles to obtain perspective images of each angle.

[0069] In the embodiments of the present invention, the detection device 108 can be a camera, a camera or an infrared ray device, etc. The detection device can be deployed in the application environment and can take pictures of the target area from multiple different angles to obtain perspective images of each angle.

[0070] The determination module includes:

[0071] The analysis sub-module is used to perform object analysis on each of the perspective images to determine the object areas where objects exist in each perspective.

[0072] The mapping sub-module is used to map the object areas where objects exist in each perspective to the reference image to determine the object areas included in the reference image.

[0073] In the embodiments of the present invention, the determination module includes an analysis sub-module and a mapping sub-module. Among them, the analysis sub-module performs object analysis on each perspective image to determine the object areas where objects exist in the perspective images of each perspective.

[0074] The mapping sub-module determines the entire object area included in the reference image by mapping the object area where an object exists at the current angle to the reference image.

[0075] The interception module includes:

[0076] The judgment sub-module is used to judge whether the object area included in the reference image is a continuous area;

[0077] The interception sub-module is used to, if the object area included in the reference image is not a continuous area, respectively intercept the images corresponding to each discrete object area included in the reference image, and use the intercepted images as the images to be identified.

[0078] In the embodiment of the present invention, the interception module includes a judgment sub-module and an interception sub-module. Among them, the judgment sub-module judges whether the object area included in the reference image is a continuous area. In the embodiment of the present invention, the objects included in the reference image may be connected together in the same area or may be scattered. The judgment sub-module determines whether the part where the object area exists is a continuous area.

[0079] The interception sub-module intercepts the image corresponding to the area containing the object. If the objects are in the same area, one image is intercepted as the image to be recognized. If it is a scattered area, one image is intercepted for each area, and the intercepted images are used as the images to be recognized.

[0080] Optionally, in the embodiment of the present invention, the intercepted images can be integrated and used as the images to be recognized.

[0081] The enhancement module includes:

[0082] An enhancement sub-module for performing enhancement processing on the image to be recognized. The enhancement processing includes at least any one or a combination of any several of grayscale conversion, binarization, denoising operation, and contrast enhancement.

[0083] In the embodiment of the present invention, the enhancement module includes an enhancement sub-module, which can perform enhancement processing on the image to be recognized and has the ability to perform operations such as grayscale conversion, binarization, denoising, and contrast enhancement.

[0084] In the embodiment of the present invention, the enhancement sub-module needs to perform any one or a combination of any several of grayscale conversion, binarization, denoising, and contrast enhancement on the image to be recognized.

[0085] The system provided by the embodiment of the present invention further includes:

[0086] A display module 109 for presenting to the user whether the recognition result of the target object output by the deep learning model is a hazardous chemical, and when the target object is a hazardous chemical, the chemical type of the target object, the hazardous chemical.

[0087] In the embodiment of the present invention, the system is provided with a display module 109 for displaying corresponding recognition information to the user.

[0088] The training module includes:

[0089] A cleaning sub-module for cleaning the multiple test images and removing the blurred images and duplicate images in the multiple test images.

[0090] In the embodiment of the present invention, a cleaning sub-module is provided in the training module, which is used to clean multiple test images and delete the blurred images and duplicate images existing in the multiple test images.

[0091] The system provided by the embodiment of the present invention further includes:

[0092] A model testing module 110, which is used to test the established deep learning model to determine the processing performance of the deep learning model.

[0093] A model optimization module 111, which is used to optimize the established deep learning model to adjust the model structure and hyperparameters of the deep learning model.

[0094] A model deployment interface 112, which is used to connect the deep learning model with a cloud server or an edge device to complete the deployment of the deep learning model in the cloud server or the edge device.

[0095] In summary, the embodiment of the present invention provides a dangerous chemical identification system based on deep learning and automatic image recognition. In practical applications, this system can inherit a variety of object detection algorithms, such as inheriting the YOLOv5 object detection algorithm, etc. Applying this dangerous chemical identification system can achieve efficient and accurate identification of various dangerous chemicals, significantly improving the detection efficiency and accuracy of dangerous chemicals.

[0096] In the dataset preparation stage of the dangerous chemical identification system provided by the embodiment of the present invention, a dataset of labeled dangerous chemical images can be used for model training to ensure that the model can identify various types of dangerous chemicals. The dataset can cover a wide range of chemical categories, such as flammable substances, explosive substances, toxic substances, etc., and strict data annotation and cleaning are carried out.

[0097] During the model training process, a variety of model algorithms can be adopted. For example, a pre-trained YOLOv5 model can be integrated and fine-tuned using a dangerous chemical dataset to adapt to specific identification tasks. The YOLOv5 model adopts a lightweight CSPDarknet53 backbone network and a PANet feature fusion method, which can effectively extract feature information in images and improve the detection accuracy and speed. The system provided by the embodiment of the present invention can select a specific model algorithm for training the deep learning model according to the actual application scenario to be identified.

[0098] In the system provided by the embodiment of the present invention, the established deep learning model can be deployed in an inference engine to achieve real-time or offline identification of dangerous chemicals. The inference engine supports obtaining image data from various security inspection devices, surveillance cameras, etc., and quickly outputs detection results, including information such as the category, location, and confidence of dangerous chemicals.

[0099] The system provided by the embodiments of the present invention can provide an intuitive and friendly user interface, facilitating users to operate and view the recognition results. The user interface supports multiple modes such as image detection, video detection, and real-time camera detection, and can be flexibly configured according to actual needs.

[0100] In the actual application process, when using the above YOLOv5 for training a deep learning model, the following application examples can be provided:

[0101] Collect and organize an image dataset containing 10 categories of common hazardous chemicals, with a total of 1288 images, and perform fine annotation. The annotation format is a txt file in YOLO format for convenient use in model training.

[0102] Use the Python language, OpenCV library, and PyTorch framework to train the YOLOv5 model in combination with the above dataset. During the training process, data augmentation techniques are adopted to enhance the generalization ability of the model, and the model parameters are optimized through multiple iterations until an ideal recognition effect is achieved.

[0103] Deploy the trained model in an inference engine and integrate it into security inspection equipment or a monitoring system. Verify the recognition accuracy and real-time performance of the system through actual tests to ensure meeting the application requirements.

[0104] Design and implement a user interface based on Qt or Web, supporting functions such as image, video, and real-time camera detection. Users can select the detection mode, upload image or video files through the interface, and view the recognition results output by the system.

[0105] Before training the model, the hazardous chemical recognition system provided by the embodiments of the present invention based on deep learning and automatic image recognition and classification can collect a large amount of image data of hazardous chemicals, and these images should cover hazardous chemicals of different types and different forms (such as liquid, solid, gas containers).

[0106] The data sources can include laboratories, industrial production sites, warehouses, transport vehicles, etc.

[0107] Clean the collected images to remove blurred, duplicate, or irrelevant images.

[0108] Perform image enhancement processing, such as grayscale conversion, binarization, denoising, contrast enhancement, etc., to improve the image quality and recognition accuracy.

[0109] Annotate the images, adding labels to each image to indicate the types of hazardous chemicals it contains.

[0110] Using deep learning techniques, such as convolutional neural network (CNN), automatically extract features related to hazardous chemicals from preprocessed images.

[0111] These features may include shape, color, texture, edges, etc.

[0112] Select a suitable deep learning model (such as YOLO, SSD, Faster R-CNN, etc.) for training.

[0113] Use the labeled image data as the training set, and continuously adjust the model parameters through forward propagation and backpropagation algorithms to optimize the recognition performance of the model.

[0114] During the training process, techniques such as cross-validation, regularization, dropout, etc. can be adopted to prevent overfitting.

[0115] Optimize the trained model, including adjusting the model structure, hyperparameters, etc., to improve the accuracy and efficiency of the model.

[0116] Methods such as ensemble learning and transfer learning can be adopted to further improve the model performance.

[0117] Use an independent test set to test the model to evaluate its performance and accuracy in practical applications.

[0118] The test set should include images of hazardous chemicals of different types and forms to comprehensively test the generalization ability of the model.

[0119] Apply the trained model to practical scenarios, such as chemical production sites, warehouses, transportation vehicles, etc.

[0120] Capture images in real time through image acquisition devices such as cameras and input them into the model for recognition.

[0121] Deploy the model to edge devices (such as cameras, smart terminals, etc.) or cloud servers.

[0122] Integrate the model with existing monitoring systems, alarm systems, etc. to achieve real-time detection and early warning of hazardous chemicals.

[0123] Continuously evaluate the performance of the model in practical applications, including indicators such as recognition accuracy and response time.

[0124] Regularly update and optimize the model to adapt to new application scenarios and requirements.

[0125] Continuously collect new image data and use it for retraining and optimization of the model.

[0126] Pay attention to new technologies and new methods in the field of deep learning and try to apply them to the field of hazardous chemical identification.

[0127] For the specific working processes of each module and sub-module in the hazardous chemical identification system based on deep learning and automatic image recognition and classification disclosed in the embodiments of the present invention above, reference may be made to the corresponding content of the hazardous chemical identification system based on deep learning and automatic image recognition and classification disclosed in the above embodiments of the present invention, and details will not be elaborated here.

[0128] Each embodiment in this specification is described in a progressive manner. For the same or similar parts among the embodiments, reference can be made to each other. Each embodiment focuses on the differences from other embodiments. In particular, for the system or system embodiments, since they are basically similar to the method embodiments, the description is relatively simple. For the relevant parts, reference can be made to the partial description of the method embodiments. The systems and system embodiments described above are merely illustrative. The units described as separate components may or may not be physically separated, and the components shown as units may or may not be physical units, that is, they may be located in one place or distributed to multiple network units. Some or all of the modules can be selected according to actual needs to achieve the purpose of the solution of this embodiment. Those of ordinary skill in the art can understand and implement it without creative efforts.

[0129] Those skilled in the art can further realize that the units and algorithm steps of each example described in combination with the embodiments disclosed in this article can be implemented by electronic hardware, computer software, or a combination of the two.

[0130] To clearly illustrate the interchangeability of hardware and software, the composition and steps of each example have been generally described according to functions in the above description. Whether these functions are executed in a hardware or software manner depends on the specific application and design constraints of the technical solution. Those skilled in the art can use different methods to implement the described functions for each specific application, but such implementation should not be considered to exceed the scope of the present invention.

[0131] The above description of the disclosed embodiments enables those skilled in the art to implement or use the present invention. Various modifications to these embodiments will be obvious to those skilled in the art. The general principles defined herein can be implemented in other embodiments without departing from the spirit or scope of the present invention. Therefore, the present invention will not be limited to these embodiments shown herein, but rather to the widest scope consistent with the principles and novel features disclosed herein.

Claims

1. A dangerous chemical identification system based on deep learning and automatic image recognition and classification, characterized in that: include: A receiving module, used to receive images of various viewing angles obtained by the detection device shooting the target area at multiple angles; A selection module, used for selecting a viewing angle image taken at a normal viewing angle as a reference image from among the viewing angle images; A determination module, used for determining an object area included in the reference image according to each of the view images; A capture module, used to capture the image where the object area is located as the image to be recognized; An enhancement module, used to perform enhancement processing on the image to be identified to obtain a target image corresponding to the image to be identified; A recognition module, used to input the target image into an established deep learning model, determine the target objects contained in the target image and the object form of each target object through processing by the deep learning model, and extract the object features of each target object according to the object form of each target object, identify whether the target object is a hazardous chemical based on the object features of each target object, and output the chemical type of the hazardous chemical to which the target object belongs when the target object is identified as a hazardous chemical; A training module, used to determine a plurality of test images, wherein the plurality of test images include images of various types of hazardous chemicals; Performing enhancement processing on the multiple test images to obtain multiple sample images; Determine each object contained in each of the sample images, and add an attribute label to each object in each of the sample images; Extracting object features of each object in each of the sample images, wherein the object features at least include the shape, color, texture and edge of the object; Determine an initial learning model, take the object features of each object in each of the sample images as input, take the attribute label of each object in each of the sample images as output, train the initial learning model until the convergence function of the initial learning model is less than a preset threshold, and complete the training of the initial learning model to establish the deep learning model; each of the attribute labels is used to characterize whether the object corresponding to it is a hazardous chemical, and when the object is a hazardous chemical, the chemical type of the hazardous chemical to which the object belongs; The detection device is used to photograph the target area from multiple angles to obtain a viewing angle image at each angle.

2. The dangerous chemical identification system based on deep learning and automatic image recognition according to claim 1 is characterized in that: The determining module comprises: An analysis submodule, configured to perform object analysis on each of the view angle images to determine an object region where an object exists at each view angle; The mapping submodule is used to map the object area where the object exists in each viewing angle to the reference image, and determine the object area contained in the reference image.

3. The dangerous chemical identification system based on deep learning and automatic image recognition according to claim 1 is characterized in that: The interception module comprises: A judging submodule, used for judging whether the object region contained in the reference image is a continuous region; The interception submodule is used to intercept the images corresponding to the discrete object regions contained in the reference image respectively if the object region contained in the reference image is not a continuous region, and use the intercepted images as the images to be identified.

4. The dangerous chemical identification system based on deep learning and automatic image recognition according to claim 1 is characterized in that: The enhancement module comprises: The enhancement submodule is used to perform enhancement processing on the image to be identified, wherein the enhancement processing at least includes any one or a combination of any several of grayscale, binarization, denoising and contrast enhancement.

5. The dangerous chemical identification system based on deep learning and automatic image recognition according to claim 1 is characterized in that: Also includes: The display module is used to display to the user the identification result of whether the target object output by the deep learning model is a hazardous chemical, and when the target object is a hazardous chemical, the chemical type of the hazardous chemical of the target object.

6. The dangerous chemical identification system based on deep learning and automatic image recognition according to claim 1 is characterized in that: The training module comprises: The cleaning submodule is used to clean the multiple test images to remove blurred images and repeated images in the multiple test images.

7. The dangerous chemical identification system based on deep learning and automatic image recognition according to claim 1 is characterized in that: Also includes: The model testing module is used to test the established deep learning model to determine the processing performance of the deep learning model.

8. The dangerous chemical identification system based on deep learning and automatic image recognition according to claim 1 is characterized in that: Also includes: The model optimization module is used to optimize the established deep learning model to adjust the model structure and hyperparameters of the deep learning model.

9. The dangerous chemical identification system based on deep learning and automatic image recognition according to claim 1 is characterized in that: Also includes: The model deployment interface is used to connect the deep learning model to the cloud server or edge device to complete the deployment of the deep learning model in the cloud server or edge device.

Images