Method and system for automatically identifying and extracting field rock core data features

Abstract

The invention discloses a field core data feature automatic identification and extraction method and system. The method comprises the following steps: S1, acquiring a field core image and preprocessing the field core image; s2, performing image labeling on the preprocessed core image to obtain image classification and core attribute description; s3, training a core target detection and classification model by using the marked image data; s4, training a rock core attribute identification model by using a rock core target detection and classification result and rock core attribute information data; and S5, based on the trained core target detection and classification model and the core attribute identification model, carrying out target detection and classification on the core image, and carrying out attribute feature identification on a detected target object. According to the method, automation and intelligentization of the whole process from data collection, preprocessing and feature extraction to model training and application are realized, manual intervention is greatly reduced, and the working efficiency is improved.

Description

Technical Field

[0001] The present invention relates to the technical field of core feature recognition and classification, and particularly to a method and system for automatically recognizing and extracting field core data features. Background Art

[0002] In the field of geological exploration, the feature recognition and classification of cores are key steps in obtaining underground structure information. Traditional core recognition methods rely on manual observation and empirical judgment, which are not only inefficient but also easily affected by human factors, resulting in limited accuracy and consistency of recognition results.

[0003] To overcome these challenges, more advanced technical means are needed in the field of geological exploration. First, geological data analysis should incorporate the latest artificial intelligence technologies, which can process more complex geological information and help engineers better understand different geological information in all aspects. Second, the application of big data analysis and deep learning algorithms can perform relevant feature recognition on geological image data and provide data support for engineering design and construction.

[0004] However, the automatic recognition and feature analysis of various geological information based on deep learning technology in the field of geological exploration are still in the initial stage. It is necessary to combine the latest technologies to automatically recognize various geological information involved in this field and obtain corresponding feature information, so as to improve its applicability and accuracy in actual projects, provide more accurate services, and meet the specific requirements under different engineering scenarios. How to combine the advantages of deep learning to construct a deep learning model capable of automatically recognizing geological images is an important challenge faced by the current field of geological image data analysis. Summary of the Invention [[ID=]18]

[0005] This application provides a method and system for automatically recognizing and extracting field core data features to solve the problem of how to improve the efficiency and accuracy of core data acquisition in geological exploration through automation technology.

[0006] According to a first aspect, in one embodiment, a method for automatically recognizing and extracting field core data features is provided. The method includes:

[0007] Step S1, collecting field core images and performing preprocessing;

[0008] Step S2, performing image annotation on the preprocessed core images to obtain image classification and core attribute description;

[0009] Step S3, training a core target detection and classification model using the annotated image data;

[0010] Step S4: Train the core attribute recognition model using the core target detection and classification results and the core attribute information data.

[0011] Step S5: Based on the trained core target detection and classification model and the core attribute recognition model, perform target detection and classification on the core image, and identify the attribute features of the detected target objects.

[0012] Further, the step S1 specifically includes:

[0013] Step S11: Obtain underground rock samples through drilling.

[0014] Step S12: Take high-definition images of the collected underground rock samples.

[0015] Step S13: Perform cropping and rotation operations on the obtained image data.

[0016] Further, the step S2 specifically includes:

[0017] Step S21: Use an image annotation tool to annotate the core target objects in the image to obtain the position information of the target objects.

[0018] Step S22: Add category information and attribute information to the core target objects in the image to generate a label file.

[0019] Further, the step S3 specifically includes:

[0020] Step S31: Divide the image data set obtained from the annotation into a training set and a validation set.

[0021] Step S32: Use the training set and the validation set to train and validate the core target detection and classification model. The input of the core target detection and classification model is the preprocessed core image, and the output is the classification detection result of the core target objects.

[0022] Further, the step S32 specifically includes:

[0023] The core target detection and classification model uses the YOLOv5 model.

[0024] Further, the step S4 specifically includes:

[0025] Step S41: Divide the image data set obtained from the target classification detection into a training set and a validation set.

[0026] Step S42: Use the training set and the validation set to train and validate the core attribute recognition model. The input of the core attribute recognition model is the core target detection object result obtained by using the core target detection and classification model, and the output is the core attribute information.

[0027] Further, step S42 specifically includes:

[0028] Train a core property recognition model for each type of property information of the core, and extract multiple types of property information through multiple parallel core property recognition models.

[0029] Further, step S42 specifically includes:

[0030] The core property recognition model uses the VGG16 model.

[0031] According to a second aspect, an embodiment provides a system for automatically recognizing and extracting field core data features, and the system includes:

[0032] An image acquisition and processing module, configured to acquire a field core image and perform preprocessing;

[0033] An image annotation module, configured to perform image annotation on the preprocessed core image to obtain image classification and core property description;

[0034] A target detection and classification model training module, configured to train a core target detection and classification model by using the annotated image data;

[0035] A property recognition model training module, configured to train a core property recognition model by using the core target detection and classification results and core property information data;

[0036] A classification and recognition module, configured to perform target detection and classification on the core image based on the trained core target detection and classification model and core property recognition model, and perform attribute feature recognition on the detected target objects.

[0037] According to a third aspect, an embodiment provides an electronic device, and the device includes: a processor and a memory;

[0038] The memory is used to store one or more program instructions;

[0039] The processor is configured to run one or more program instructions to execute the steps of a method for automatically recognizing and extracting field core data features as described in any one of the above.

[0040] The present application provides a method and system for automatically identifying and extracting characteristics of field core data. Introducing deep learning into the analysis operations in geological exploration can significantly improve the efficiency and accuracy of field data collection. Through standardized core photo shooting, automated orthorectification, cropping and annotation, and a classification model based on machine learning, the data quality, classification accuracy are improved, manual operations are reduced, and costs are lowered. In addition, this technology can be continuously optimized and iterated. Through the feedback mechanism of production personnel, new data sets are continuously collected to improve the accuracy and generalization performance of the model. Through the method of the present invention, rock characteristics can be automatically extracted, the recognition accuracy can be improved, the dependence on expert experience can be reduced, and the efficiency and accuracy of engineering exploration work can be improved, thus providing a new perspective and tool for geological research and exploration, and promoting the in-depth development of geological research and oil and gas exploration and development. BRIEF DESCRIPTION OF THE DRAWINGS

[0041] Figure 1 It is a flowchart of a method for automatically identifying and extracting characteristics of field core data provided by an embodiment of the present invention;

[0042] Figure 2 It is a specific implementation flowchart of a method for automatically identifying and extracting characteristics of field core data provided by an embodiment of the present invention;

[0043] Figure 3 It is an example of a core image in a method for automatically identifying and extracting characteristics of field core data provided by an embodiment of the present invention;

[0044] Figure 4 It is a schematic diagram of the logical structure of a system for automatically identifying and extracting characteristics of field core data provided by an embodiment of the present invention;

[0045] Figure 5 It is a schematic diagram of the physical structure of an electronic device provided by an embodiment of the present invention. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0046] The present invention will be further described in detail below in conjunction with the drawings through specific embodiments. Similar elements in different embodiments are labeled with related similar element numbers. In the following embodiments, many details are described to enable a better understanding of the present application. However, those skilled in the art can easily recognize that some of these features can be omitted in different situations, or can be replaced by other elements, materials, and methods. In some cases, some operations related to the present application are not shown or described in the specification to avoid the core part of the present application being overwhelmed by excessive description. For those skilled in the art, it is not necessary to describe these related operations in detail, and they can fully understand the related operations based on the description in the specification and general technical knowledge in the art.

[0047] In addition, the features, operations, or characteristics described in the specification can be combined in any suitable manner to form various embodiments. At the same time, the steps or actions in the method description can also be reordered or adjusted in a manner obvious to those skilled in the art. Therefore, the various sequences in the specification and drawings are only for clearly describing a certain embodiment and do not mean a necessary sequence, unless it is stated that a certain sequence must be followed.

[0048] A method for automatically identifying and extracting the characteristics of field core data provided by the first embodiment of the present invention will be described in detail below in conjunction with Figure 1 and Figure 2 for detailed description.

[0049] As Figure 1 shown, in step S1, a field core image is collected and preprocessed.

[0050] Collection of core image data: First, core photos are taken in the field according to standard regulations, including appropriate lighting, angles, and resolutions, and these photos and related formation stratification record information are collected.

[0051] The above steps specifically include:

[0052] Step S11, obtaining underground rock samples through drilling;

[0053] Step S12, taking high-definition images of the collected underground rock samples;

[0054] Step S13, performing cropping and rotation operations on the obtained image data.

[0055] Specifically, in this step, when working in the field, the staff uses drilling equipment to drill into the ground to obtain samples of underground rocks. During the drilling process, the drill bit drills into the underground rocks to form cylindrical core samples, as Figure 3 shown. Then, during the digital imaging acquisition of the core, to ensure the clarity of the core photos and the authenticity of the colors, the acquisition process needs to use a front vertical view as the common shooting angle and shoot the core under uniform and soft natural light or artificial light source. In addition, the pixel resolution of the core photos should be high enough to ensure that the tiny details and texture features in the photos can be clearly captured. It is recommended to use a camera with at least more than 5 million pixels for shooting to obtain high-definition images. In addition, it should be ensured that the resolution of the captured image is sufficient to support enlarged viewing without distortion. Usually, it is recommended to use a resolution of 300 dpi (dots per inch) for printing-quality photo shooting. Finally, the obtained image data is cropped into a 640*640 picture. Secondly, the obtained image is rotated to obtain positive image data.

[0056] AsFigure 1 As shown, in step S2, image annotation is performed on the preprocessed core image to obtain image classification and core attribute description.

[0057] Data annotation: Use an annotation tool to perform detailed annotation on each core photo, including project name, borehole number, depth range, geographical location coordinates, core name, and physical and chemical property descriptions of the rock and soil. At the same time, standardize and store this information to provide high-quality annotated data for subsequent deep learning model training. This step is a key link to ensure that the model can effectively learn and identify core features.

[0058] The above steps specifically include:

[0059] Step S21, use an image annotation tool to annotate the core target object in the image to obtain the position information of the target object;

[0060] Step S22, add category information and attribute information to the core target object in the image to generate a label file.

[0061] Specifically, in this step, use the image annotation tool LabelImg to perform precise classification and attribute description on the core image. According to the actual situation of the image, it is mainly divided into different types of cores (for example, soil layer, diorite, etc.). The attributes are artificially defined, mainly for the attributes of different types of cores, such as color, hardness, integrity, etc., and generate a label file in YOLO format. This label file records all different category labels in the dataset. Each line represents a category name. For each picture, a.txt file with the same name as it will be generated to store the annotation information of all target objects in the picture. The content of this file consists of multiple lines, and each line represents the annotation information of a target object. At the same time, organize the pictures and label files into a certain directory structure for easy reading during training. For example, a dataset root directory containing two subdirectories, images and labels, can be created, where the images directory stores the picture files and the labels directory stores the corresponding label files. Finally, a dataset configuration file (such as.yaml format) is required, which specifies the paths of the training, validation, and test data, as well as the category information, etc.

[0062] .txt file contains the annotation information of all target objects in this image. This information is used to train the YOLO model so that it can identify and locate different objects in the image. For example <class_index><x_center><y_center> <width> <height>Each row represents the annotation information of a target object, specifically including:

[0063] <class_index>: The class index of the target object. This is an integer starting from 0, indicating the class to which the target object belongs. For example, if there are two classes: "rock" and "soil", and "rock" is assigned the class index 0, "soil" is assigned the class index 1, then the class index of a target object belonging to the "rock" class will be 0.

[0064] <x_center> and <y_center>: The center point coordinates of the target object in the image. These coordinates are proportional values relative to the image width and height, ranging from 0 to 1. For example, if the center of the target object is located exactly in the middle of the image, then both <x_center> and <y_center> will be 0.5.

[0065] <width>and <height>: The width and height of the target object. These values are also proportional values relative to the image width and height, ranging from 0 to 1.

[0066] As Figure 1 shown, in step S3, the core target detection and classification model is trained using the labeled image data.

[0067] Recognition of specific objects in core images: Apply the YOLOv5 model to core images to obtain the regions of interest in the core image data, extract key visual features for classification and recognition, such as shape, texture, color, and spatial relationships, etc. These features are then used to build an accurate target detection and classification model, thereby achieving efficient analysis of various geological features in core images.

[0068] The above steps specifically include:

[0069] Step S31, divide the training set and the validation set based on the labeled image dataset;

[0070] Step S32, use the training set and the validation set to train and validate the core target detection and classification model. The input of the core target detection and classification model is the preprocessed core image, and the output is the classification detection result of the core target object. In this embodiment, the core target detection and classification model uses the YOLOv5 model.

[0071] Specifically, in this step, the pictures and label files of the dataset prepared in S2 are used, and the data is divided into a training set and a validation set at the same time. Secondly, modify the YOLOv5 data configuration file, specify the path of the dataset, the number of categories, and the category names. According to needs, modify the model configuration file (.yaml). In the YOLOv5 configuration file, set the depth_multiple parameter to 0.5, modify the width_multiple to 0.6, the value of nc to 10, the anchors to [30,61,62,45,59,119], use the lr-find command to automatically find the best initial learning rate, and set the weight_decay to 0.0005.

[0072] In this embodiment, the input of the YOLOv5 model is: pre-processed image data and a.txt file with the same name as the image generated by the annotation software. The output is the detection results (categories) of all target objects in the image and the corresponding position information (bounding box information). YOLOv5 is a popular and efficient object detection model, featuring high speed, ease of use, and high flexibility. YOLOv5 and VGG16 are two different types of deep learning models. In this patent, YOLOv5 is used to identify the categories of core samples, and VGG16 is used to further analyze the attributes of these features, such as the density and hardness of rocks. The combined use of these two models can provide more comprehensive geological data analysis. In this way, YOLOv5 and VGG16 can work together to provide a complete solution from object detection to attribute analysis.

[0073] The YOLOv5 data configuration file (usually a.yaml file) is used to configure and define the dataset. It tells the model where to find the training and validation data and how to properly process this data. It mainly includes specifying the path of the data, defining the categories, setting the dataset division, category names, and indices.

[0074] As Figure 1 shown, in step S4, the core property recognition model is trained using the core object detection and classification results and the core property information data.

[0075] Recognition of core image attribute features in the target area: For the target core image data extracted in the above steps, the data is divided into a training set and a test set, and multiple deep learning models based on VGG16 are trained to automatically recognize the relevant attributes of the target core image, such as the density category of soil, the hardness, integrity, and weathering degree of rocks. Finally, the models for multiple attribute classifications are fused into a complete automatic recognition system for core data attributes.

[0076] The above steps specifically include:

[0077] Step S41, dividing the training set and the validation set based on the image dataset obtained from the target classification detection;

[0078] Step S42, using the training set and the validation set to train and validate the core property recognition model. The input of the core property recognition model is the core object detection result obtained by the core object detection and classification model, and the output is the core property information.

[0079] In this embodiment, a core property recognition model is trained for each type of core property information, and multiple parallel core property recognition models are used to extract multiple types of property information.

[0080] In this embodiment, the core attribute recognition model adopts the VGG16 model.

[0081] Specifically, in this step, the target image dataset obtained in S3 is divided into an 80% training set and a 20% test set. Secondly, according to the attributes of the target image to be recognized, the corresponding deep learning model is selected. In this patent, for each attribute (such as density, hardness, integrity, weathering degree, etc.), multiple independent VGG16 models are trained. Specifically, first, the image is preprocessed, adjusted to 224x224 pixels, and the pre-trained vgg16 model is used as the initial starting point for model training. That is, most of the weights of the pre-trained model are frozen, and only the last few layers (usually the fully connected layers) are trained. Set appropriate learning rates and optimizers (such as Adam), and use the cross-entropy loss function for multi-classification tasks (each attribute includes multiple classifications, for example, the hardness is divided into soft, hard, and medium). Finally, integrate all systems: integrate each independent model into a unified system to ensure that they can work together. Specifically, obtain the prediction results through each independent VGG16 model in turn. Multiple VGG16 models work in parallel to process the above image data at the same time, and finally integrate these results and return them.

[0082] In this embodiment, the steps of selecting data for each attribute and training a VGG16 model, for example, through yolov5, it is recognized that the core belongs to diorite:

[0083] 1. Dataset Division

[0084] Divide the labeled dataset into a training set and a test set. Usually, the training set is used to train the model, and the test set is used to evaluate the performance of the model.

[0085] 2. Attribute Data Selection

[0086] For each attribute, select relevant images and annotation information from the labeled data. For example, if there is an attribute "rock density", then images marked with density information and the corresponding annotation files need to be selected from all labeled data.

[0087] 3. Data Preprocessing

[0088] Preprocess the selected image data to meet the input requirements of the VGG16 model. VGG16 usually requires input images of a fixed size, so the image may need to be cropped or resized. In addition, normalization processing may also be required to enable the model to learn better.

[0089] 4. Train the VGG16 Model

[0090] Train the VGG16 model using the selected attribute data. VGG16 is a pre-trained deep convolutional neural network that can be used for image classification tasks. During training, the model will learn to recognize features related to the selected attributes.

[0091] In this embodiment, the method of "freezing most of the weights of the pre-trained model and only training the last few layers" is a common transfer learning strategy. The core idea of this method is to use the model pre-trained on a large dataset to extract features, while fine-tuning some layers of the model for a specific task. In transfer learning, it is generally believed that the higher layers (close to the output layer) of the pre-trained model can better capture task-specific features. Therefore, we only train these layers so that the model can adapt to the new, specific task.

[0092] When using the Keras library in Python, the weights of the model can be frozen by setting the trainable attribute. The following are the specific implementation steps:

[0093] Load the pre-trained model: Load a pre-trained model, such as VGG16.

[0094] Freeze the weights: Set the trainable attribute of all layers in the model except the last few layers to False.

[0095] Add custom layers: Add new layers at the end of the model, and these layers will be trained to adapt to the new task.

[0096] Compile the model: Compile the model to prepare for training.

[0097] Train the model: Train the model, and only the weights of the newly added layers and the last few layers will be updated.

[0098] The formula for the cross-entropy loss function is as follows:

[0099]

[0100] Where: y i is an indicator variable. If class i is the true class, then y i = 1, otherwise y i = 0. p i is the probability that the model predicts that the sample belongs to class i. C is the total number of classes. In a multi-class classification problem, the cross-entropy loss function calculates the weighted sum of the logarithms of the probabilities of all classes.

[0101] As Figure 1 shown, in step S5, based on the trained core target detection and classification model and the core attribute recognition model, perform target detection and classification on the core image, and perform attribute feature recognition on the detected target objects.

[0102] Model Loading and Service Provision: Load the core automatic recognition and feature extraction system, put the model into production, and provide a feedback interface for production personnel. If the answer is incorrect, the production personnel can directly give the correct answer. The model improves the accuracy and generalization performance by continuously collecting data sets and iterating.

[0103] Specifically, in this step, based on the integrated system model in S4, this part loads the model training files from the perspective of online use and then conducts relevant application-level operations on the model. Integrate the model into a predetermined production environment, including web services. At the same time, integrate a feedback function that allows users to provide correct answers when the recognition results are inaccurate, and regularly analyze user feedback to understand the performance of the model service in actual applications and user needs.

[0104] The present invention provides a method for automatic recognition and analysis of field core images integrated with deep learning technology. This method has revolutionarily improved the traditional core recognition process in the field of geological exploration, greatly enhancing the automation processing level and recognition accuracy of core data. First, by formulating a series of strict data collection standards, including light, angle, and resolution, etc., the acquisition quality of field core images is ensured. At the same time, through a detailed image annotation process, key information including project name, borehole number, depth range, geographical location coordinates, core name, and descriptions of geotechnical physical and chemical properties is collected, laying a solid foundation for establishing a high-quality training data set. Secondly, advanced image processing techniques are used to preprocess the collected core photos, including angle correction, de-distortion, cropping, and enhancement, etc., to eliminate shooting errors and redundant image information, ensuring the quality and consistency of image data and providing accurate inputs for the training of deep learning models. Then, deep learning technologies such as YOLOv5 model and VGG16 are used to extract features and classify core images, automatically identifying geological features in core images, such as color, texture, structure, etc., improving the accuracy and efficiency of core image recognition. Finally, aiming at the complexity that core images may contain multiple geological features, the present invention adopts multi-label image recognition technology, which can simultaneously identify multiple geological features in the image, enhancing the comprehensiveness and accuracy of recognition. At the same time, a feedback mechanism for production personnel is constructed, enabling the model to iterate and optimize according to user inputs, and continuously improving the accuracy and generalization ability of the model.

[0105] The present invention realizes the full - process automation and intelligence from data collection, pre - processing, feature extraction to model training and application, greatly reducing manual intervention and improving work efficiency. The implementation of the present invention not only realizes the automation and intelligence of core image recognition technically, but also has broad application prospects and important practical value in practical applications. It is expected to provide strong technical support for the intelligent transformation of the geological exploration industry, with broad application prospects and practical value.

[0106] Corresponding to the above - disclosed method for automatically identifying and extracting field core data features, an embodiment of the present invention also discloses a system for automatically identifying and extracting field core data features, as Figure 4 shown, which specifically includes:

[0107] An image acquisition and processing module, configured to acquire field core images and perform pre - processing;

[0108] An image annotation module, configured to perform image annotation on the pre - processed core images to obtain image classification and core attribute description;

[0109] A target detection and classification model training module, configured to train a core target detection and classification model using the annotated image data;

[0110] An attribute recognition model training module, configured to train a core attribute recognition model using the core target detection and classification results and core attribute information data;

[0111] A classification and recognition module, configured to perform target detection and classification on core images based on the trained core target detection and classification model and core attribute recognition model, and perform attribute feature recognition on the detected target objects.

[0112] It should be noted that for the detailed description of a system for automatically identifying and extracting field core data features provided in an embodiment of the present invention, reference can be made to the relevant description of a method for automatically identifying and extracting field core data features provided in an embodiment of the present application, which will not be elaborated here.

[0113] In another embodiment, an electronic device is provided. Figure 5 Schematic diagram of the physical structure of the electronic device provided by an embodiment of the present invention. The electronic device may include: a processor 301, a communications interface 302, a memory 303, and a bus 304. Among them, the processor 301, the communications interface 302, and the memory 303 communicate with each other through the bus 304. The processor 301 can call a computer program stored in the memory 303 and executable on the processor 301 to execute the method for automatically identifying and extracting field core data features provided in the above embodiment.

[0114] In addition, when the logical instructions in the above-mentioned memory 303 are implemented in the form of software function units and sold or used as independent products, they can be stored in a computer-readable storage medium. Based on such an understanding, the technical solution of the embodiment of the present invention, in essence, or the part that contributes to the prior art, or a part of this technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions for causing a computer device (which may be a personal computer, a server, or a network device, etc.) to execute all or part of the steps of the methods described in various embodiments of the present invention. The foregoing storage medium includes: various media such as USB flash drives, mobile hard disks, read-only memories (ROM, Read-Only Memory), random access memories (RAM, Random Access Memory), magnetic disks, or optical discs that can store program codes.

[0115] In another embodiment, a computer storage medium is provided. The storage medium can be located in any one of the computer terminals in a computer terminal group in a computer network, or in any one of the mobile terminals in a mobile terminal group.

[0116] Optionally, in this embodiment, the storage medium is set to store program codes for executing the steps of the method for automatically identifying and extracting field core data features provided in the above embodiment.

[0117] Those skilled in the art can understand that all or part of the functions of the various methods in the above embodiments can be implemented in a hardware manner or in a computer program manner. When all or part of the functions in the above embodiments are implemented in a computer program manner, the program can be stored in a computer-readable storage medium. The storage medium can include: read-only memory, random access memory, magnetic disk, optical disk, hard disk, etc. The above functions can be realized by a computer executing the program. For example, storing the program in the memory of the device, when the processor executes the program in the memory, the above all or part of the functions can be realized. In addition, when all or part of the functions in the above embodiments are implemented in a computer program manner, the program can also be stored in a storage medium such as a server, another computer, magnetic disk, optical disk, flash drive or mobile hard disk, and saved to the memory of the local device by downloading or copying, or the system of the local device can be updated in version. When the processor executes the program in the memory, all or part of the functions in the above embodiments can be realized.

[0118] The above uses specific examples to elaborate on the present invention, which is only used to help understand the present invention and is not intended to limit the present invention. For the technology to which the present invention belongs.< / height> < / width> < / height> < / width>

Claims

1. An automatic recognition and extraction method for the characteristics of field core data, characterized in that, The method includes: Step S1, collecting field core images and performing preprocessing; Step S2, performing image annotation on the preprocessed core images to obtain image classification and core attribute description; Step S3, training a core object detection and classification model using the annotated image data; Step S4, training a core attribute recognition model using the core object detection and classification results and core attribute information data; Step S5, based on the trained core object detection and classification model and core attribute recognition model, performing object detection and classification on the core images, and performing attribute feature recognition on the detected target objects.

2. The automatic recognition and extraction method of field core data features according to claim 1, characterized in that The step S1 specifically includes: Step S11, obtaining underground rock samples through drilling; Step S12, taking high-definition images of the collected underground rock samples; Step S13, performing cropping and rotation operations on the obtained image data.

3. The automatic recognition and extraction method for the characteristics of field core data according to claim 1, characterized in that, The step S2 specifically includes: Step S21, using an image annotation tool to annotate the core target objects in the image to obtain the position information of the target objects; Step S22, adding category information and attribute information to the core target objects in the image to generate a label file.

4. The automatic recognition and extraction method for field core data features according to claim 1, characterized in that, The step S3 specifically includes: Step S31, dividing the image data set obtained based on the annotation into a training set and a validation set; Step S32, using the training set and the validation set to train and validate the core object detection and classification model. The input of the core object detection and classification model is the preprocessed core image, and the output is the classification detection result of the core target objects.

5. The automatic recognition and extraction method of field core data features according to claim 4, characterized in that, The step S32 specifically includes: The core object detection and classification model adopts the YOLOv5 model.

6. The automatic recognition and extraction method for characteristics of field core data according to claim 1, characterized in that, The step S4 specifically includes: Step S41, dividing the image data set obtained based on the target classification detection into a training set and a validation set; Step S42, using the training set and the validation set to train and validate the core attribute recognition model. The input of the core attribute recognition model is the core object detection result obtained using the core object detection and classification model, and the output is the core attribute information.

7. The automatic recognition and extraction method for field core data features according to claim 6, characterized in that, The step S42 specifically includes: Training a core attribute recognition model for each type of attribute information of the core, and extracting multiple types of attribute information through multiple parallel core attribute recognition models.

8. The automatic recognition and extraction method for characteristics of field core data according to claim 7, characterized in that, The step S42 specifically includes: The core attribute recognition model adopts the VGG16 model.

9. An automatic recognition and extraction system for the characteristics of field core data, characterized in that, The system includes: An image acquisition and processing module for collecting field core images and performing preprocessing; An image annotation module for performing image annotation on the preprocessed core images to obtain image classification and core attribute description; A target detection and classification model training module for training a core object detection and classification model using the annotated image data; An attribute recognition model training module for training a core attribute recognition model using the core object detection and classification results and core attribute information data; A classification and recognition module for performing object detection and classification on the core images based on the trained core object detection and classification model and core attribute recognition model, and performing attribute feature recognition on the detected target objects.

10. An electronic device, characterized in that, The device includes: a processor and a memory; The memory is used for storing one or more program instructions; The processor is used for running one or more program instructions to execute the steps of a method for automatically identifying and extracting field core data features according to any one of claims 1 to 8.

Images