Moon dark mantle sediment identification method and system based on yov8 model

Abstract

The invention discloses a moon dark mantle sediment identification method and system based on a yov8 model, and the method comprises the steps: introducing a multi-scale feature extraction module into a moon dark mantle sediment identification model in a process that a skeleton network module outputs a feature map, enabling the model to employ a multi-scale convolutional neural network to extract the features of an image, and carrying out the recognition of the features of the image; the scale adaptability and nonlinearity of the model are enhanced, and the performance of detecting the multi-scale target by the model is improved. According to the method, the dark mantle sediment data and a deep learning algorithm are combined, so that the recognition capability on complex terrains and sediments with different scales is improved. Meanwhile, the method has higher detection precision and speed, can quickly respond to requirements under different observation conditions, and provides more reliable data support for scientific research. The method provided by the invention can provide important scientific basis for detection of the sediment of the dark mantle, astronomical exploration, mineral resource evaluation, revealing of the volcanic activity history and the structural evolution process of the early stage of the moon and the like.

Description

Technical Field

[0001] The invention belongs to the technical field of deep space exploration, and specifically relates to a remote sensing image processing technology based on a yolov8 model, which is used for a method of identifying dark mantle deposits (DMD) on the lunar surface. Background Art

[0002] Dark Mantle Deposits (DMD) are one of the important geological features formed by volcanic eruptions on the lunar surface. DMD areas are usually rich in elements such as iron and titanium. Previous workers divided the deposits into two groups: "regional" dark mantle deposits (RDMD), which are characterized by thin and flat cover deposits covering an area of ​​about thousands of square kilometers, and "localized" dark mantle deposits (LDMD), which are smaller in area (usually less than 100 square kilometers) and have recognizable positive topographic features. These deposits are usually dark in color and are mainly composed of glassy volcanic eruptions. They were mainly formed during the early lunar volcanic activity. The ejected materials were cooled and deposited on the lunar surface, covering a part of the younger basalt terrain. The study of DMD helps to reveal the history of early lunar volcanic activity and tectonic evolution. Traditionally, the identification of DMD relies on manual analysis of high-resolution remote sensing images. Researchers usually visually identify dark areas in images and combine spectral and topographic data to confirm whether they are DMDs. However, this method is inefficient, time-consuming, and relies on the professional experience of researchers, and is easily affected by subjective factors. In addition, manual analysis is difficult to ensure the comprehensiveness and accuracy of identification when faced with massive amounts of lunar data. With the development of artificial intelligence and deep learning technology, computer vision algorithms are increasingly used in remote sensing image analysis. Convolutional neural networks (CNNs) have made significant progress in tasks such as target detection and image classification, greatly improving the efficiency of image processing.

[0003] Although deep learning models have performed well in the field of ground object detection, their application to DMD recognition in lunar remote sensing images still faces many challenges. First, the lighting conditions and reflectivity characteristics of the lunar surface are complex, and there is a lot of noise in the images, which may affect the detection accuracy of the model. Second, the distribution of DMD is complex and easily confused with other lunar landforms. Traditional automated recognition algorithms perform poorly on such tasks.

[0004] The YOLO (You Only Look Once) model is a representative algorithm in the field of target detection in recent years. Its unique architecture enables it to have real-time detection capabilities while maintaining high detection accuracy. YOLOv8 (You Only Look Once Version 8) is a target detection model based on deep learning. It performs well in real-time object recognition and positioning tasks. The model achieves efficient feature extraction and fast reasoning through a fully convolutional neural network (CNN) architecture. YOLOv8 has been trained on multiple data sets and can accurately detect and classify a variety of targets. However, the model still faces many challenges in applying it to DMD recognition in lunar remote sensing images. Summary of the invention

[0005] In order to solve the problems of efficiency and reliability in the identification of lunar dark mantle deposits (DMD), the present invention provides a lunar dark mantle deposit identification method and system based on the YOLOv8 model, which uses deep learning technology to achieve efficient identification of DMD, and verifies the model prediction results in combination with the lunar derived FeO and TiO2 percentage content data, thereby improving the reliability and scientificity of the identification results.

[0006] According to one aspect of the present invention, a method for identifying lunar dark mantle sediments based on the YOLOv8 model is provided, comprising: Obtain slice images of lunar remote sensing image data; Inputting the slice image into a trained lunar dark mantle sediment recognition model to output a predicted potential DMD area; Use FeO and TiO2 data to verify potential DMD areas, identify potential DMD areas that meet preset requirements and mark them in lunar remote sensing images; The training of the lunar dark mantle sediment identification model includes: Construct a lunar remote sensing image dataset with DMD regions labeled; Construct the yolov8 model and embed a multi-scale feature extraction module in the model's skeleton network to accurately focus on areas with high DMD correlation; Based on the constructed DMD dataset, the yolov8 model is trained using an adaptive optimizer and a dynamic batch size strategy, and the trained model is output as a lunar dark mantle sediment recognition model.

[0007] As a further technical solution, a lunar remote sensing image dataset with DMD region annotated was constructed, including: Obtain remote sensing image data of the lunar surface and obtain DMD position coordinates in DMD-related papers; The obtained DMD position coordinates are marked on the lunar global map to establish a DMD data set.

[0008] As a further technical solution, based on the constructed DMD dataset, the yolov8 model is trained using an adaptive optimizer and a dynamic batch size strategy, including: The skeleton network is used to extract features from the input DMD training samples and output feature maps. The multi-scale feature extraction module is used to extract multi-scale features from the output feature maps. The feature fusion network is used to sample and fuse the extracted multi-scale features. The head output module is used to output the prediction box.

[0009] As a further technical solution, the multi-scale feature extraction module includes two branches, one branch uses residual connection to speed up model training; and the other branch uses channel attention mechanism, average pooling and spatially separable convolution to extract DMD feature information.

[0010] As a further technical solution, the slice image is input into a trained lunar dark mantle sediment recognition model to output a predicted potential DMD area, and further includes: Multiple detection boxes are generated according to feature maps of different scales, and the potential DMD area with the highest confidence is output based on the detection threshold.

[0011] As a further technical solution, the FeO and TiO2 data were used to verify the potential DMD areas, including: The lunar FeO and TiO2 data are compared with the predicted potential DMD areas to generate the geographical location of the DMD areas that meet the preset requirements, as well as the matching information between the DMD areas and the FeO and TiO2 percentage content data.

[0012] According to one aspect of the present invention, a lunar dark mantle sediment identification system based on the yolov8 model is provided, comprising: A data acquisition module, used to obtain slice images of lunar remote sensing image data; A potential DMD region prediction module is used to input the slice image into a trained lunar dark mantle sediment recognition model and output a predicted potential DMD region; The DMD region verification and identification module is used to verify the potential DMD region using FeO and TiO2 data, identify the potential DMD region that meets the preset requirements and mark it in the lunar remote sensing image; The training of the lunar dark mantle sediment identification model includes: Construct a lunar remote sensing image dataset with DMD regions labeled; Construct the yolov8 model and embed a multi-scale feature extraction module in the model's skeleton network to accurately focus on areas with high DMD correlation; Based on the constructed DMD dataset, the yolov8 model is trained using an adaptive optimizer and a dynamic batch size strategy, and the trained model is output as a lunar dark mantle sediment recognition model.

[0013] According to one aspect of the invention specification, a device for identifying lunar dark mantle sediments based on the YOLOv8 model is provided, comprising a memory and a processor, wherein the memory stores program instructions executed by the processor, and the processor calls the program instructions to execute the steps of the method for identifying lunar dark mantle sediments based on the YOLOv8 model.

[0014] According to one aspect of the present specification, a non-transitory computer-readable storage medium is provided, wherein the non-transitory computer-readable storage medium stores computer instructions, wherein the computer instructions enable the computer to execute the steps of the method for identifying lunar dark mantle sediments based on the YOLOv8 model.

[0015] Compared with the prior art, the present invention has the following beneficial effects: 1. The present invention uses the yolov8 model to achieve efficient DMD recognition: the yolov8 model can maintain efficient target detection performance when processing large-scale remote sensing images, and its fast feature extraction and detection capabilities are particularly suitable for DMD recognition tasks. Even under the conditions of a small data set for the lunar dark mantle sediment recognition scene, by introducing a multi-scale feature extraction module in the yolov8 model, the model's ability to extract DMD detail features at different scales is enhanced.

[0016] 2. The present invention verifies the prediction results by combining FeO and TiO2 data. The lunar derived FeO and TiO2 percentage content data are important chemical composition indicators in lunar geology. Since DMD areas are usually rich in elements such as iron and titanium, they can help distinguish DMD areas. In DMD identification, the use of derived FeO and TiO2 data to verify the DMD areas predicted by the model can effectively improve the reliability of the model in identification. Through this verification method, it is ensured that the DMD areas identified by the model are consistent with the actual geological characteristics. BRIEF DESCRIPTION OF THE DRAWINGS

[0017] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, a brief introduction is given below to the drawings used in the embodiments or the description of the prior art. Obviously, the drawings described below are some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying any creative work.

[0018] Figure 1A schematic flow chart of a method for identifying lunar dark mantle sediments based on the YOLOv8 model provided in an embodiment of the present invention.

[0019] Figure 2 A schematic diagram of the training process of the lunar dark mantle sediment identification model provided in an embodiment of the present invention.

[0020] Figure 3 A schematic diagram of the multi-scale feature extraction module flow provided in an embodiment of the present invention.

[0021] Figure 4 A schematic diagram of the process of predicting potential DMD areas of dark mantle sediments using a model provided in an embodiment of the present invention.

[0022] Figure 5 Schematic diagram of dark mantle deposits provided in an embodiment of the present invention, the left picture is a regional dark mantle deposit, and the right picture is a local dark mantle deposit.

[0023] Figure 6 A schematic diagram of some effects of the lunar dark mantle sediment identification process provided by an embodiment of the present invention. DETAILED DESCRIPTION

[0024] The terms "including" and "having" and any variations thereof in the specification and claims of the present invention and the above-mentioned drawings are intended to cover non-exclusive inclusions, for example, a process, method, system, product or apparatus comprising a series of steps or units is not necessarily limited to the steps or units explicitly listed, but may include other steps or units not explicitly listed or inherent to these processes, methods, products or apparatuses.

[0025] It should be noted that: The YOLOv8 model performs well in the field of ground object detection, but its application to DMD recognition in lunar remote sensing images still faces many challenges. First, the lighting conditions and reflectivity characteristics of the lunar surface are complex, and there is a lot of noise in the image, which may affect the detection accuracy of the model. Secondly, the distribution of DMD is complex and easily confused with other lunar landforms. Traditional automated recognition algorithms perform poorly on such tasks. Therefore, a new method that combines deep learning models and geological verification data is needed to improve recognition efficiency while ensuring the scientific rationality of the detection results.

[0026] To solve the above problems, the present invention proposes a DMD identification method based on the yolov8 model by introducing a multi-scale feature extraction (MSFE) module, and combines the derived FeO and TiO2 percentage content data to verify the test results to ensure the accuracy and scientificity of the identification. This method can not only achieve efficient DMD identification, but also improve the reliability of the identification results by comparing the physical and chemical components.

[0027] In order to make the purpose, technical solutions and advantages of the embodiments of the present invention clearer, the technical solutions in the embodiments of the present invention will be clearly and completely described below in conjunction with the drawings in the embodiments of the present invention. Obviously, the described embodiments are part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments in the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention. In addition, the technical features in the various embodiments or single embodiments provided by the present invention are arbitrarily combined with each other to form a new technical solution. This combination is not restricted by the sequence of steps and / or the structural composition mode, but must be based on the ability of ordinary technicians in this field to achieve. When the combination of technical solutions is contradictory or cannot be achieved, it should be considered that this combination of technical solutions does not exist and is not within the scope of protection required by the present invention.

[0028] See also Figure 1 The lunar dark mantle sediment identification method based on the yolov8 model provided in an embodiment of the present invention includes: obtaining a slice image of lunar remote sensing image data; inputting the slice image into a trained lunar dark mantle sediment identification model, and outputting a predicted potential DMD area; verifying the potential DMD area using FeO and TiO2 data, identifying the potential DMD area that meets the preset requirements and marking it in the lunar remote sensing image.

[0029] The method described in the embodiment of the present invention introduces a multi-scale feature extraction module based on the yolov8 model to identify lunar dark mantle sediments, aiming to use deep learning technology to achieve efficient identification of DMD, and verify the model prediction results in combination with lunar-derived FeO and TiO2 percentage content data to improve the reliability and scientificity of the identification results. The method of the present invention can not only cope with the limitations of small data sets, but also ensure the reliability of the identification results by comparing chemical composition data.

[0030] See also Figures 2 to 4 The construction, training and identification process of the lunar dark mantle sediment identification model described in the embodiment of the present invention includes the following steps: Step 1, download the data and obtain the DMD position coordinates in the DMD related papers.

[0031] Obtain high-resolution remote sensing image data of the lunar surface, including data from missions such as the Lunar Reconnaissance Orbiter (LRO) and Clementine UVVIS. Specifically: Lunar Reconnaissance Orbiter Camera (LROC) Wide Angle Camera (WAC) global morphology map, used to establish a DMD dataset for the acquired DMD coordinates; Clementine UVVIS-derived FeO abundance map and Lunar Reconnaissance Orbiter Camera (LROC) Wide Angle Camera (WAC)-derived TiO2 abundance map, used for subsequent result verification to ensure the reliability of model predictions. Obtain DMD position coordinates in DMD-related papers.

[0032] Step 2: Mark the obtained DMD position coordinates on the lunar global map and establish a DMD dataset.

[0033] In step 2, the lunar optical remote sensing image is cut into training samples of appropriate size, and the DMD data set is manually labeled using the labeling tool (LabelImg). The core task of labeling is to frame the DMD area with a rectangular frame and add a "DMD" label to each frame to ensure that the labeling file format is correct and the labeling coordinates are accurate.

[0034] Specifically, the obtained DMD position coordinates are marked on the downloaded Lunar Reconnaissance Orbiter Camera (LROC) Wide Angle Camera (WAC) global morphology map, and the DMD area is clipped using the ARCGIS raster clipping tool to obtain the DMD dataset. It should be noted that the dataset should include accurate annotations of the DMD area so that the model can learn to recognize features.

[0035] Step 3: construct a lunar dark mantle sediment identification network model and train it on the DMD dataset; The yolov8 model is trained using the annotated DMD dataset. When the dataset is fed into the model, the model scales the image to adjust its resolution (usually 640x640 pixels) to meet the model's input specifications. This step helps improve the efficiency and effectiveness of model training. The performance of the model is then optimized by introducing the MSFE module. During the training process, the model's loss function is monitored to ensure that the model can maintain a high recognition accuracy on the validation set.

[0036] The lunar dark mantle sediment recognition model in step 3 consists of five parts: input, skeleton network (Backbone), multi-scale feature extraction module (MSFE), feature fusion network (Neck) and detection head (Head). During the model training process, the input image is first processed and feature extracted by the skeleton network, and a powerful convolutional neural network CSPNet (Cross Stage Partial Network) is used as the skeleton network. The backbone network extracts multi-level features from the input image through a series of convolution operations, including: low-level features (Low-Level Features) are used to capture basic information such as edges, textures and colors in the image and high-level features (High-Level Features) are used to capture complex geometric shapes and structural information of objects. After that, the feature map output by the skeleton network is extracted from multiple scales through the multi-scale feature extraction module to ensure that the model can detect objects of various sizes. Then, the neck performs feature fusion. The model combines the feature pyramid network (FPN) and the path aggregation network (PAN). Through the fusion of multi-level feature maps, the detection ability of targets of different scales is enhanced. This part enables the model to simultaneously detect smaller detail targets and larger global structure targets in the image. Finally, the head output is used to predict the object. The detection head of the model processes each feature map in the feature pyramid and outputs the prediction result. The main task of the detection head is to regress the bounding box position of each candidate box and predict the category and confidence score of the object in each candidate box. After the candidate box prediction is completed, the model removes highly overlapping candidate boxes through NMS and only retains the bounding box with the highest confidence.

[0037] This application innovatively introduces the multi-scale feature extraction (MSFE) module into the YOLOv8 model. Although there have been studies on similar introductions in the YOLOv5 model, the two models face different technical situations due to differences in architectural characteristics. YOLOv5 focuses on rapid detection of basic targets, and the introduction of MSFE focuses on lightweight adaptation; while YOLOv8 is oriented to complex high-resolution images (which fits remote sensing image tasks), and the integration of MSFE requires overcoming a series of difficulties. First, it is necessary to fine-tune the interface and dimensionality rules between the multi-scale feature extraction module and the native network layer to prevent feature mismatch and information flow barriers; second, for the computing power and memory mode of YOLOv8, it is necessary to readjust the hyperparameter settings. For example, the ECA part needs to optimize the adaptive kernel parameters based on the new channel dimension to ensure efficient use of computing resources; third, for FeO and TiO2 data, prediction verification is performed after the model training is completed, which further improves the detection accuracy of the YOLOv8 model for remote sensing image tasks. By combining these data to verify the prediction results, the carefully designed MSFE module effectively improves the performance of the model in high-precision detection, achieving a significant improvement in detection accuracy over the basic framework.

[0038] In the present invention, the multi-scale feature extraction (MSFE) module is introduced into the YOLOv8 model. We analyze in detail the difference between its integration mode and the MSFE integration mode in the YOLOv5 model, and make corresponding adjustments to ensure that it can efficiently adapt to the complex network architecture of YOLOv8. Although both YOLOv5 and YOLOv8 use the MSFE module for feature extraction, due to the differences in the network architecture between the two, the integration mode and optimization strategy of the MSFE module are also different. YOLOv5 is mainly used for rapid detection of targets, and its architecture is relatively simple. Therefore, the MSFE module is embedded in the middle layer of the network as a feature fusion module to extract and fuse multi-scale features. The focus is on quickly extracting and fusing features of different scales to meet the requirements of lightweight tasks. YOLOv8 is a more complex model with a deeper and more detailed network structure, which is suitable for processing high-resolution and complex remote sensing image data. In YOLOv8, the MSFE module is embedded in the deep structure of the skeleton network, thereby enhancing the model's ability to extract and recognize fine-grained features. The introduction of the MSFE module requires that its output feature map can be correctly connected to other layers in the YOLOv8 network. Therefore, during the integration process, its input and output dimensions must be precisely adjusted to ensure that they match the input and output dimensions of each layer of the YOLOv8 network. Specifically, we need to adjust the adaptation layer to ensure that the output dimension of the MSFE module matches the input dimension of the subsequent network layer. By precisely controlling the number of channels, width and height dimensions of the feature map, we can effectively avoid feature mismatches and information flow barriers, and ensure that information in each layer of the network can be smoothly transmitted.

[0039] The multi-scale feature extraction module (MSFE) flexibly configures submodules according to task characteristics and model architecture differences, aiming to accurately capture multi-scale features of images. It has a dual-branch structure: the first branch uses residual connection, which is a classic architecture optimization method in the field of deep learning. It effectively alleviates gradient disappearance with the help of jump connection, participates in strict data flow control through adaptive adjustment, deeply fits the overall architecture, and realizes model training acceleration; the second branch integrates channel attention mechanism (ECA), average pooling and spatial convolution layer, and works together to efficiently extract DMD feature information, which complements the first branch and fully enables MSFE's excellent performance.

[0040] The multi-scale feature extraction module consists of channel attention mechanism (ECA), average pooling and spatially separable convolution. The ECA module adaptively adjusts the convolution kernel size based on the feature vector calculated by global average pooling (GAP). This process needs to be adjusted in YOLOv8 based on the new channel dimension to ensure that the convolution kernel can achieve the maximum effect on larger or smaller channel dimensions. It is necessary to ensure that the kernel size matches the complexity of the network to avoid too large or too small convolution kernels affecting the feature extraction ability. The channel attention mechanism (ECA) focuses on improving feature selection capabilities without adding too much computational burden. ECA uses the global average pooling method without dimensionality reduction to aggregate features, adaptively determines the kernel size k, and then performs a one-dimensional 1D convolution of the kernel size k on the feature map, and then obtains the channel attention vector through the Sigmoid function. ECA is calculated as follows: (1) (2) Where σ is the Sigmoid function, Conv1D is a 1-D convolution with a kernel size of k, and the kernel size k is adaptively determined according to the channel dimension C. GAP(X) represents the global average pooling operation, which is the result of global average pooling of each channel of the input feature map, converted into a channel-level statistic vector. W represents the channel attention vector. X and Y represent the input feature map and the output feature map, respectively.

[0041] (3) where |x| odd represents the nearest odd number to x, γ and b are hyperparameters, represents a function mapping relationship that generates a specific output value according to the input channel dimension C. We assign the values ​​of formulas (2) and (1) to γ ​​and b.

[0042] Average pooling takes the average value of pixels in each corresponding area of ​​the input feature map according to the given kernel size to complete the downsampling operation. Because there are no parameters to be optimized, it will not over-learn the details of the training data, thus avoiding overfitting. The calculation method of average pooling is as follows: (4) In the formula, s represents the kernel size, Indicates the points in the calculation area The pixel value of .

[0043] Spatially separable convolution: Spatially separable convolution splits a standard convolution operation into multiple small kernel convolution operations in the spatial dimension.

[0044] (5) (6) in Indicates the midpoint of k×k convolution The value of represents the value of the point (i,1) in the k-times ×1 convolution, Represents the value of the midpoint (1,j) of the 1×k convolution. The number of parameters for k×1 convolution and 1×k convolution is (k+1)+(1+k)=2k, which is smaller than the number of parameters for k×k convolution when k>2. Spatially separable convolution reduces module parameters, speeds up model calculation, and increases model depth.

[0045] When the multi-scale feature extraction (MSFE) module is integrated into the YOLOv8 model, due to the complexity of its network structure, the hyperparameters in the training process need to be fine-tuned, especially the learning rate and batch size. First, the optimization of the learning rate is crucial. A high learning rate may cause gradient explosion, while a too low learning rate will slow down the training process. To this end, a learning rate scheduling strategy can be adopted, step decay, and the learning rate can be gradually reduced to help the model converge stably. In addition, the adaptive optimizer (Ranger) can automatically adjust the learning rate according to the gradient, improve training stability, and solve the computational complexity problem introduced by the MSFE module. Secondly, the batch size also needs to be optimized. A larger batch size helps improve computational efficiency, reduce the fluctuation of gradient estimation, and improve training stability. However, this will also increase the demand for video memory, so it is necessary to find a balance between memory and computing power. A dynamic batch size strategy can be adopted to use a smaller batch size for training in the early stage, and gradually increase the batch size as the training progresses, thereby accelerating the training. By fine-tuning these hyperparameters, the MSFE module can be stably trained in YOLOv8 and improve model performance.

[0046] Step 4: Use the model trained in step 3 to predict and verify DMD for the lunar global slice image.

[0047] First, the global morphological map of the Lunar Reconnaissance Orbiter Camera (LROC) Wide Angle Camera (WAC) is sliced, and then the model is loaded using the saved model weights to maintain the network structure and parameters during training. The sliced ​​image is input into the model, and the forward propagation network part of the model is used to generate the prediction result. After the prediction result is generated, some post-processing steps are required to perform non-maximum suppression (NMS) to filter out the detection frames with too much overlap, so as to obtain the potential DMD detection result, and finally the predicted potential DMD area is identified and detected. The results output by the model will be presented in the form of a detection frame, marking the potential DMD area. In the present invention, for the identification of lunar dark mantle deposits (DMD), the content range of FeO (wt.%) and TiO2 (wt.%) is specified as FeO: 10-17.5 wt.%, TiO2: 2-4.5 wt.%. The selection of this range is based on the analysis of DMD characteristics in different regions of the moon and the research results of relevant literature. In the present invention, the content range of FeO and TiO2 is mainly used to screen possible DMD areas, and combined with other features of remote sensing images for comprehensive judgment to improve the recognition accuracy of the model for DMD. In addition, FeO and TiO2 data are used to verify the prediction results of the model. By comparing the differences in FeO and TiO2 content in the predicted area with the geological characteristics of the background area, the consistency of the test results with the actual geological characteristics is ensured.

[0048] Step 5, find the longitude and latitude coordinates of the predicted DMD obtained in step 4, and mark the potential DMD that meets the requirements on the global remote sensing image of the moon. Figure 5 A schematic diagram of dark mantle deposits is given, where the left picture shows regional dark mantle deposits and the right picture shows local dark mantle deposits. Figure 6 A schematic diagram of some of the effects of the lunar dark mantle sediment identification process is given.

[0049] The implementation basis of each embodiment of the present invention is to implement programmed processing through a device with a processor function. Therefore, in engineering practice, the technical solutions and functions of each embodiment of the present invention are encapsulated into various modules. Based on this reality, on the basis of the above embodiments, an embodiment of the present invention provides a lunar dark mantle sediment identification system based on the yolov8 model, which is used to execute the lunar dark mantle sediment identification method based on the yolov8 model in the above method embodiment.

[0050] The system comprises: a data acquisition module for acquiring slice images of lunar remote sensing image data; a potential DMD region prediction module for inputting the slice images into a trained lunar dark mantle sediment recognition model and outputting predicted potential DMD regions; a DMD region verification and recognition module for verifying potential DMD regions using FeO and TiO2 data, identifying potential DMD regions that meet preset requirements and marking them in lunar remote sensing images; wherein the training of the lunar dark mantle sediment recognition model comprises: constructing a lunar remote sensing image data set with DMD regions annotated; constructing a yolov8 model, and embedding a multi-scale feature extraction module in the skeleton network of the model to accurately focus on regions with high DMD correlation; based on the constructed DMD data set, the yolov8 model is trained using an adaptive optimizer and a dynamic batch size strategy, and the trained model is output as a lunar dark mantle sediment recognition model.

[0051] The lunar dark mantle sediment identification system based on the yolov8 model provided by the embodiment of the present invention is aimed at the current situation that "the existing technology mainly relies on manual analysis of remote sensing images, which is inefficient and easily affected by subjective factors, and it is difficult to process a large amount of image data; and deep learning technology is prone to overfitting when training deep learning models on small data sets, and the model performs poorly when processing complex geological features". The above-mentioned several modules are adopted to realize efficient identification of DMD by using deep learning technology, and the model prediction results are verified in combination with the lunar derived FeO and TiO2 percentage content data, so as to improve the reliability and scientificity of the identification results and solve the problems of efficiency and reliability in the identification of lunar dark mantle sediments.

[0052] It should be noted that the system embodiments provided by the present invention are not only used to implement the methods in the above-mentioned method embodiments, but also used to implement the methods in other method embodiments provided by the present invention. The only difference lies in the setting of corresponding functional modules, and the principles thereof are basically the same as the principles of the above-mentioned system embodiments provided by the present invention. As long as technical personnel in this field refer to the specific technical solutions in other method embodiments on the basis of the above-mentioned system embodiments, obtain corresponding technical means and technical solutions composed of these technical means by combining technical features, and on the premise of ensuring the practicality of the technical solutions, they will improve the modules in the above-mentioned system embodiments to obtain corresponding system class embodiments for implementing the methods in other method class embodiments.

[0053] Based on the same inventive concept as the above-mentioned embodiment, an embodiment of the present invention also provides a lunar dark mantle sediment identification device based on the YOLOv8 model, comprising a memory and a processor, wherein the memory stores program instructions executed by the processor, and the processor calls the program instructions to execute the steps of the lunar dark mantle sediment identification method based on the YOLOv8 model.

[0054] The lunar dark mantle sediment identification device based on the yolov8 model in the above embodiment introduces a multi-scale feature extraction module based on the yolov8 model for DMD identification, and verifies the detection results in combination with the derived FeO and TiO2 percentage content data to ensure the accuracy and scientificity of the identification. The above device can not only realize efficient DMD identification, but also improve the reliability of the identification results by comparing the physical and chemical components.

[0055] Based on the same inventive concept as the above-mentioned embodiment, an embodiment of the present invention also provides a non-transitory computer-readable storage medium, which stores computer instructions, and the computer instructions enable the computer to execute the steps of the lunar dark mantle sediment identification method based on the YOLOv8 model.

[0056] In summary, the present invention discloses a method and system for identifying lunar dark mantle sediments based on multi-source remote sensing data and the YOLOv8 model. When training the lunar dark mantle sediment identification model based on the YOLOv8 model, a multi-scale feature extraction module is introduced in the process of outputting the feature map of the skeleton network module, so that the model uses a multi-scale convolutional neural network to extract the features of the image, effectively enhancing the scale adaptability and nonlinearity of the model, and improving the performance of the model in detecting multi-scale targets. Compared with the prior art, the innovation of the present invention is to combine the dark mantle sediment data with the deep learning algorithm to improve the recognition ability of complex terrain and sediments of different scales. At the same time, the method has higher detection accuracy and speed, can quickly respond to the needs under different observation conditions, and provide more reliable data support for scientific research. After the detection of dark mantle sediments, it can provide important scientific basis for astronomical exploration, mineral resource assessment, and revealing the history of volcanic activity and tectonic evolution process of the early moon.

[0057] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention, rather than to limit it. Although the present invention has been described in detail with reference to the aforementioned embodiments, those skilled in the art should understand that they can still modify the technical solutions described in the aforementioned embodiments, or replace some or all of the technical features therein with equivalents. However, these modifications or replacements do not deviate the essence of the corresponding technical solutions from the technical solutions of the embodiments of the present invention.

Claims

1. A lunar dark mantle sediment identification method based on the yolov8 model, characterized in that: include: Obtain slice images of lunar remote sensing image data; Inputting the slice image into a trained lunar dark mantle sediment recognition model to output a predicted potential DMD area; Use FeO and TiO2 data to verify potential DMD areas, identify potential DMD areas that meet preset requirements and mark them in lunar remote sensing images; The training of the lunar dark mantle sediment identification model includes: Construct a lunar remote sensing image dataset with DMD regions labeled; Construct the yolov8 model and embed a multi-scale feature extraction module in the model's skeleton network to accurately focus on areas with high DMD correlation; Based on the constructed DMD dataset, the yolov8 model is trained using an adaptive optimizer and a dynamic batch size strategy, and the trained model is output as a lunar dark mantle sediment recognition model.

2. The method for identifying lunar dark mantle sediments based on the yolov8 model according to claim 1, characterized in that: Construct a lunar remote sensing image dataset with DMD area annotated, including: Obtain remote sensing image data of the lunar surface and obtain DMD position coordinates in DMD-related papers; The obtained DMD position coordinates are marked on the lunar global map to establish a DMD data set.

3. The method for identifying lunar dark mantle sediments based on the yolov8 model according to claim 1 is characterized in that: Based on the constructed DMD dataset, the yolov8 model is trained using an adaptive optimizer and a dynamic batch size strategy, including: The skeleton network is used to extract features from the input DMD training samples and output feature maps. The multi-scale feature extraction module is used to extract multi-scale features from the output feature maps. The feature fusion network is used to sample and fuse the extracted multi-scale features. The head output module is used to output the prediction box.

4. The method for identifying lunar dark mantle sediments based on the yolov8 model according to claim 3 is characterized in that: The multi-scale feature extraction module includes two branches, one branch adopts residual connection to speed up the model training speed; the other branch adopts channel attention mechanism, average pooling and spatially separable convolution to extract DMD feature information.

5. The method for identifying lunar dark mantle sediments based on the yolov8 model according to claim 1, characterized in that: Inputting the slice image into the trained lunar dark mantle sediment identification model to output the predicted potential DMD area also includes: Multiple detection boxes are generated according to feature maps of different scales, and the potential DMD area with the highest confidence is output based on the detection threshold.

6. The method for identifying lunar dark mantle sediments based on the yolov8 model according to claim 1, characterized in that: Validation of potential DMD regions using FeO and TiO2 data also includes: The lunar FeO and TiO2 data are compared with the predicted potential DMD areas to generate the geographical location of the DMD areas that meet the preset requirements, as well as the matching information between the DMD areas and the FeO and TiO2 percentage content data.

7. The lunar dark mantle sediment identification system based on the yolov8 model is characterized by: include: A data acquisition module, used to obtain slice images of lunar remote sensing image data; A potential DMD region prediction module is used to input the slice image into a trained lunar dark mantle sediment recognition model and output a predicted potential DMD region; The DMD region verification and identification module is used to verify the potential DMD region using FeO and TiO2 data, identify the potential DMD region that meets the preset requirements and mark it in the lunar remote sensing image; The training of the lunar dark mantle sediment identification model includes: Construct a lunar remote sensing image dataset with DMD regions labeled; Construct the yolov8 model and embed a multi-scale feature extraction module in the model's skeleton network to accurately focus on areas with high DMD correlation; Based on the constructed DMD dataset, the yolov8 model is trained using an adaptive optimizer and a dynamic batch size strategy, and the trained model is output as a lunar dark mantle sediment recognition model.

8. A lunar dark mantle sediment identification device based on the yolov8 model, characterized in that: It includes a memory and a processor, the memory stores program instructions executed by the processor, and the processor calls the program instructions to execute the steps of the lunar dark mantle sediment identification method based on the YOLOv8 model as described in any one of claims 1 to 6.

9. A non-transitory computer-readable storage medium, characterized in that: The non-transitory computer-readable storage medium stores computer instructions, which enable the computer to execute the steps of the lunar dark mantle sediment identification method based on the YOLOv8 model as described in any one of claims 1 to 6.

Images