Pipeline shallow profile image automatic identification system based on deep learning

Abstract

The invention discloses a deep learning-based pipeline shallow profile image automatic identification system and method. The system comprises an introduction module, a model improvement module, a data acquisition module, a training module and an identification module. The model improvement module integrates an attention feature pyramid network, a target detection dynamic head structure and an optimization loss function into the basic model to form an initial recognition model. The attention feature pyramid network dynamically adjusts the feature map extraction weight; the target detection dynamic head structure adjusts parameters according to the image complexity; and optimizing penalty terms of the loss function adjustment angle, scale and length-width ratio. The data acquisition module performs acquisition to form a training set, a test set and a verification set. The training module trains the initial model based on the training set data, and adjusts the hyper-parameter through the verification set until the model performance meets a preset condition. And the identification module identifies the test set image by using the optimization model to obtain a final result. According to the invention, the automatic identification precision and efficiency of the pipeline shallow profile image can be effectively improved.

Description

Technical Field

[0001] The present invention relates to the technical field of image recognition, and in particular to an automatic recognition system for pipeline shallow section images based on deep learning. Background Art

[0002] As an important infrastructure in marine engineering, submarine pipelines are widely used in the fields of oil and gas transportation, submarine communications, resource development, and environmental monitoring. Due to the particularity of the submarine environment, the safety monitoring and maintenance of pipelines have become particularly important, involving regular inspections of pipelines, fault detection, and environmental adaptability assessments. With the continuous development of submarine engineering technology, the monitoring method of pipelines has gradually shifted from traditional manual inspections and physical detection methods to more efficient and accurate automated technologies, and image recognition technology has played an important role in this.

[0003] Traditional submarine pipeline detection methods usually rely on image processing-based technologies, especially when using ground penetrating radar (GPR) for submarine pipeline detection, image recognition technology has become one of the core means. At present, advanced image recognition models based on YOLOv5 algorithm, dual YOLOv8-pose model and YOLOv8n algorithm are widely used in this field. YOLO series algorithms have achieved good results due to their efficient target detection capabilities, especially in land environments, and can quickly detect and locate targets. In particular, YOLOv5 and YOLOv8 series algorithms have optimized feature extraction and target detection accuracy through deep learning models, and have played an important role in practical applications.

[0004] However, although algorithms such as YOLOv5 and YOLOv8 perform well in traditional land environments, their application in complex submarine environments faces many challenges. Shallow images in submarine pipeline detection usually present complex background noise due to the particularity of the underwater environment, and the pipeline targets are small in size and varied in shape. The complex underwater background and uneven illumination in deep waters make it difficult for existing target detection technologies to cope with these challenges. The recognition accuracy and efficiency of submarine pipelines are difficult to meet actual needs, especially the recognition accuracy of small-sized targets and the processing of background noise, which is still a major bottleneck of current technology.

[0005] In response to this problem, existing technologies have gradually begun to adopt the YOLOv8 algorithm for improvement, in order to improve the accuracy and efficiency of pipeline identification in more complex submarine environments. The YOLOv8 algorithm has multiple technical advantages over previous models. First, through the improved feature fusion mechanism (FPN+PAN), YOLOv8 can effectively process multi-scale features, thereby more accurately detecting small targets in the submarine environment. Secondly, the dynamic anchor frame mechanism adopted by the YOLOv8 algorithm further improves its adaptability in complex backgrounds, enabling it to more effectively identify pipeline targets of different scales in complex submarine scenes. In addition, the advanced data enhancement technology and dynamic label allocation strategy of the YOLOv8 algorithm further improve the generalization ability and detection accuracy of the model, significantly reducing false detection and missed detection.

[0006] Despite this, the application of the YOLOv8 algorithm in submarine pipeline detection still faces certain limitations. For example, in low-resolution images, the detailed information of pipeline targets is easily lost, which affects the detection effect. Moreover, factors such as complex background and uneven lighting in the submarine environment often lead to interference from similar background objects, increasing the risk of false detection. In addition, although YOLOv8 has improved in recognition accuracy, its computational requirements are high, which may limit its application in scenarios with high real-time requirements or limited hardware resources. Finally, YOLOv8's ability to handle target occlusion is still insufficient, and its performance is more dependent on the diversity and quality of training data, which also makes the performance of the model in complex environments uncertain.

[0007] Therefore, although the YOLOv8 algorithm has shown great potential in submarine pipeline detection, in order to further improve its performance in complex submarine environments, it is still necessary to improve the accuracy and efficiency of submarine pipeline detection in complex submarine environments. Summary of the invention

[0008] In view of the shortcomings of the prior art, the purpose of the present invention is to provide a pipeline shallow section image automatic recognition system based on deep learning, which is used to improve the accuracy and efficiency of submarine pipeline detection in complex submarine environments.

[0009] To achieve the above object, the present invention provides the following technical solution: a pipeline shallow section image automatic recognition system based on deep learning, comprising:

[0010] The import module is used to introduce the YOLOv8 model as the basic model;

[0011] A model improvement module, connected to the introduction module, for integrating the attention feature pyramid network, the target detection dynamic head structure and the optimization loss function into the basic model to form an initial recognition model;

[0012] The attention feature pyramid network dynamically adjusts the weight of feature extraction according to different feature maps, the target detection dynamic head structure automatically adjusts the parameter configuration of the detection head according to the feature complexity of the input image, and the optimization loss function dynamically adjusts the penalty terms of angle, scale and aspect ratio to optimize the direction error of the detection frame;

[0013] A data acquisition module is used to collect multiple submarine pipeline shallow profile images, annotate them to form a pipeline shallow profile image data set, and then divide the pipeline shallow profile image data set into a training set, a test set and a verification set;

[0014] A training module, connected to the data acquisition module and the model improvement module, respectively, for training the initial recognition model according to the data in the training set, and adjusting hyperparameters in the initial recognition model training process according to the data in the validation set, until the performance parameters of the model meet the preset parameter conditions, and outputting the model as an optimized recognition model;

[0015] The recognition module is connected to the training module and the data acquisition module, and is used to recognize the image data in the test set according to the optimized recognition model to obtain the pipeline shallow section image recognition result.

[0016] Furthermore, the optimization loss function is a SIoU function, and the SIoU function is also used to dynamically adjust the penalty term of the center point distance. The penalty term calculation formula of the center point distance is configured as:

[0017] ,

[0018] in, A penalty term for the distance from the center point, and They are used to represent the central horizontal coordinate and central vertical coordinate of the prediction box of the target detection task, and They are used to represent the central horizontal coordinate and central vertical coordinate of the real box of the target detection task respectively.

[0019] Furthermore, the penalty calculation formula for the scale and aspect ratio is configured as:

[0020] ,

[0021] in, A penalty term for the scale and aspect ratio, and are used to represent the width and height of the prediction box respectively, and are used to represent the width and height of the real frame respectively;

[0022] The penalty calculation formula for the angle is configured as:

[0023] ,

[0024] in, A penalty term for the angle, and are respectively used to represent the angle between the predicted frame and the real frame;

[0025] The formula configuration of the SIoU function is:

[0026] ,

[0027] in, Used to represent the SIoU function, Used to represent the traditional IoU function, , and They are respectively used to represent the preset first weight coefficient, second weight coefficient and third weight coefficient.

[0028] Furthermore, the ratio of data in the training set, the test set and the validation set is 10:5:1.

[0029] Furthermore, the performance parameters include accuracy, recall, average precision mean, model parameter quantity and total floating-point operations.

[0030] Furthermore, it also includes an image processing module, which is connected to the data acquisition module and is used to perform background separation, image denoising and target enhancement on the original image data in the training set, the test set and the validation set in sequence to obtain optimized image data.

[0031] Furthermore, it also includes an environment detection module connected to the training module, and the environment detection module is used to detect the light brightness and light uniformity of the seabed environment where the pipeline is located, and detect the relative distance from the pipeline;

[0032] The training module inputs the light illuminance, the light uniformity, and the relative distance into a preset learning rate optimization formula to calculate an optimized learning rate, and inputs the light illuminance, the light uniformity, and the relative distance into a preset weight optimization formula to calculate an optimized weight attenuation parameter;

[0033] The optimized recognition model improves the model structure according to the optimized learning rate and the optimized weight decay parameter and then outputs it.

[0034] Furthermore, the learning rate optimization formula is configured as:

[0035] ,

[0036] in, is used to denote the optimization learning rate, It is used to represent the initial learning rate of the optimized recognition model, It is used to represent the adjustment coefficient of the light brightness to the learning rate, Used to indicate the brightness of the light. An exponential adjustment factor for representing the effect of the light intensity on the learning rate, Used to express the relative distance influence coefficient, Used to indicate the uniformity of illumination, Used to indicate the relative distance, Used to represent the distance factor adjustment coefficient.

[0037] Furthermore, the weight optimization formula is configured as:

[0038] ,

[0039] in, Used to represent the optimization weight decay coefficient, is used to represent the initial weight decay coefficient of the optimized recognition model, It is used to represent the adjustment coefficient of the light brightness to the weight attenuation. It is used to indicate the influence index of the light brightness on the attenuation parameter. An adjustment factor used to indicate the uniformity of illumination. It is used to represent the attenuation parameter adjustment coefficient related to the pipeline distance. Used to indicate the preset time decay coefficient, Used to represent time variables, An adjustment factor used to represent the effect of time factors on attenuation.

[0040] A pipeline shallow section image automatic recognition method based on deep learning, applied to the above-mentioned pipeline shallow section image automatic recognition system based on deep learning, comprising:

[0041] Step S1, the introduction module introduces the YOLOv8 model as the basic model;

[0042] Step S2, the model improvement module integrates the attention feature pyramid network, the target detection dynamic head structure and the optimization loss function into the basic model to form an initial recognition model, the attention feature pyramid network dynamically adjusts the weight of feature extraction according to different feature maps, the target detection dynamic head structure automatically adjusts the parameter configuration of the detection head according to the feature complexity of the input image, and the optimization loss function dynamically adjusts the penalty items of angle, scale and aspect ratio to optimize the direction error of the detection frame;

[0043] Step S3, the data acquisition module acquires a plurality of submarine pipeline shallow profile images, and annotates them to form a pipeline shallow profile image dataset, and then divides the pipeline shallow profile image dataset into a training set, a test set, and a verification set;

[0044] Step S4, the training module trains the initial recognition model according to the data in the training set, and adjusts the hyperparameters in the initial recognition model training process according to the data in the validation set, until the performance parameters of the model meet the preset parameter conditions, and outputs it as an optimized recognition model;

[0045] Step S5, the recognition module recognizes the image data in the test set according to the optimized recognition model to obtain a pipeline shallow section image recognition result.

[0046] Beneficial effects of the present invention:

[0047] The present invention improves the YOLOv8 model by integrating optimization technologies such as attention feature pyramid network, target detection dynamic head structure and optimized loss function, and combines high-quality data collection and training mechanism to train the optimized recognition model, which significantly improves the target detection accuracy, stability and efficiency of submarine pipeline shallow section images. Compared with traditional detection methods, the present invention can achieve more efficient and accurate target detection in complex submarine environments, and has good real-time performance and application value. BRIEF DESCRIPTION OF THE DRAWINGS

[0048] Figure 1 It is a structural schematic diagram of the pipeline shallow section image automatic recognition system in the present invention;

[0049] Figure 2 It is a model structure diagram of the optimized recognition model in the present invention;

[0050] Figure 3 is the image to be detected in Embodiment 1 of the present invention;

[0051] Figure 4 is the image detected by the YOLOv8 original model in Example 1 of the present invention;

[0052] Figure 5 is an image detected by the optimized recognition model in Example 1 of the present invention;

[0053] Figure 6 It is a flow chart of the steps of the pipeline shallow section image automatic recognition method in the present invention.

[0054] Figure numerals: 1. Introduction module; 2. Model improvement module; 3. Data acquisition module; 4. Training module; 5. Recognition module; 6. Image processing module; 7. Environment detection module. DETAILED DESCRIPTION

[0055] The present invention is further described in detail below in conjunction with the accompanying drawings and embodiments. The same parts are represented by the same reference numerals. It should be noted that the words "front", "rear", "left", "right", "upper" and "lower" used in the following description refer to directions in the accompanying drawings, and the words "bottom surface" and "top surface", "inner" and "outer" refer to directions toward or away from the geometric center of a specific component, respectively.

[0056] Example 1, reference Figure 1 , which is the first embodiment of the present invention, provides a pipeline shallow section image automatic recognition system based on deep learning, which can improve the accuracy and efficiency of submarine pipeline detection in complex submarine environments, including:

[0057] Module 1 is introduced to introduce the YOLOv8 model as the basic model;

[0058] Model improvement module 2, connected to introduction module 1, is used to integrate the attention feature pyramid network, target detection dynamic head structure and optimized loss function into the basic model to form an initial recognition model;

[0059] The attention feature pyramid network dynamically adjusts the weights of feature extraction according to different feature maps. The target detection dynamic head structure automatically adjusts the parameter configuration of the detection head according to the feature complexity of the input image. The optimization loss function dynamically adjusts the penalty terms of angle, scale and aspect ratio to optimize the directional error of the detection frame.

[0060] The data acquisition module 3 is used to collect multiple submarine pipeline shallow profile images, annotate them to form a pipeline shallow profile image dataset, and then divide the pipeline shallow profile image dataset into a training set, a test set and a verification set;

[0061] The training module 4 is connected to the data acquisition module 3 and the model improvement module 2 respectively, and is used to train the initial recognition model according to the data in the training set, and adjust the hyperparameters in the initial recognition model training process according to the data in the validation set, until the performance parameters of the model meet the preset parameter conditions and output it as the optimized recognition model, wherein the performance parameters include accuracy, recall rate, average precision mean, model parameter quantity and total floating-point operation quantity;

[0062] The recognition module 5 is connected to the training module 4 and the data acquisition module 3, and is used to recognize the image data in the test set according to the optimized recognition model to obtain the pipeline shallow section image recognition result.

[0063] Working principle of embodiment 1:

[0064] Module 1 is used to introduce YOLOv8 as the basic model for target detection, providing structural support for subsequent improvements. By loading the pre-trained YOLOv8 model weights, the model has preliminary target detection capabilities.

[0065] Model improvement module 2 integrates the attention feature pyramid network, target detection dynamic head structure and optimization loss function based on the YOLOv8 model to form an initial recognition model.

[0066] Among them, the attention feature pyramid network (AFPN) optimizes feature extraction at different scales by introducing an adaptive feature fusion mechanism, and enhances the detection capability of small targets (such as pipeline fractures, foreign objects, etc.). As an adaptive feature fusion structure, it has played a significant role in improving the performance of YOLOv8. AFPN introduces an adaptive weight mechanism based on the feature pyramid network, which can dynamically adjust the weights according to the feature maps of different levels, so that the network can extract target features more accurately at different scales. Compared with the traditional FPN (Feature Pyramid Network), AFPN can more effectively solve the problem of feature redundancy and insufficiency in multi-scale target detection, especially when detecting small targets. Secondly, AFPN further enhances the complementarity between features and the efficiency of information transmission through the design of cross-scale information flow. This not only improves the network's target detection capability in complex scenes, but also reduces the occurrence of false detection and missed detection, especially when dealing with challenging scenes such as occlusion and complex background. By integrating AFPN in the YOLOv8 model, experiments show that the mean average precision (mAP) and detection speed of the network have been improved. Especially in the evaluation of pipeline shallow profile image datasets, the YOLOv8 model enhanced by AFPN showed higher accuracy and better generalization ability compared with the traditional FPN structure.

[0067] Object Detection Dynamic Head Structure (LADH): A lightweight adaptive detection head is used to dynamically adjust the parameter configuration according to the complexity of the input image to reduce the consumption of computing resources and improve the detection speed. LADH (Lightweight AdaptiveDynamic Head) introduces a more flexible and efficient object detection head structure, which greatly optimizes the performance of the model. The core advantage of LADH lies in its lightweight design and adaptive characteristics, which enables the network to significantly improve the detection speed and resource utilization efficiency while maintaining high accuracy. First, LADH adopts a dynamic adjustment strategy to automatically adjust the parameter configuration of the detection head according to the feature complexity of the input image. This adaptive mechanism enables the network to dynamically adjust the allocation of computing resources according to different input scenarios, especially when dealing with complex scenes or extreme target distributions, it can more efficiently extract and utilize key features. This dynamic structure greatly reduces the computational overhead and improves the inference efficiency. Secondly, LADH reduces the excessive redundant calculations in the traditional detection head through lightweight design while maintaining sufficient expressive power. It uses modularization and weight sharing strategies to reduce model parameters while ensuring sensitivity and accuracy to target features. Experiments show that LADH significantly reduces the model parameters and computational complexity of YOLOv8, and demonstrates superior performance in edge devices and real-time applications.

[0068] By integrating LADH, the YOLOv8 model has demonstrated excellent results on multiple standard datasets. In the evaluation of pipeline shallow section image datasets, the LADH structure significantly shortens the model's inference time while maintaining a high level of mean average precision (mAP). Compared with the static head structure, LADH significantly improves the detection capability of the YOLOv8 model in multi-target and multi-scale scenarios, and significantly improves the detection accuracy of small targets.

[0069] SIoU (optimization loss function) optimizes the matching of target boxes, reduces the error of detection boxes, and improves the robustness of target detection by dynamically adjusting the penalty items of angle, scale, and aspect ratio.

[0070] Data acquisition module 3 collects images of measured data of shallow profile of pipelines in a certain area on the seabed, including 163 shallow profile images of submarine pipelines with pipeline features at different angles and sizes, and forms a pipeline shallow profile image data set through the image annotation tool LabelImg, and divides the data set into training set, test set, and validation set for model training, testing and tuning, where the ratio of data in the training set, test set, and validation set is 10:5:1. Table 1 is a table of data proportions of the training set, test set, and validation set. The data proportions of the training set, test set, and validation set are shown in Table 1:

[0071] Table 1

[0072]

[0073] In Table 1, the number of image data in the training set, test set, and validation set are 100, 50, and 10, respectively.

[0074] Training module 4 uses the data in the training set to train the initial recognition model, and adjusts the hyperparameters (such as learning rate, batch size, optimizer type, weight decay parameters, etc.) during the training process through the data in the validation set to improve the model performance. During the training process, the Adam optimizer is used for gradient update, and the model parameters are gradually optimized in combination with the cosine annealing learning rate scheduling.

[0075] Among them, Table 2 is the experimental environment parameter table, and the experimental environment parameters are shown in Table 2:

[0076] Table 2

[0077]

[0078] In Example 1, the initial learning rate of the training is set to 0.01, the initial weight decay parameter is 0.0005, the batch sample size is 64, the number of training rounds is 200, and the input image size is 640×640×3.

[0079] When the mean average precision (mAP) on the validation set reaches a preset threshold (such as above 85%), the optimized recognition model is output.

[0080] The recognition module 5 uses the optimized recognition model in the test set to calculate the accuracy, recall, mAP and other indicators to verify the model performance. The recognition result output is the target detection box coordinates, category label, and confidence score for subsequent applications such as submarine pipeline health monitoring and defect identification.

[0081] The model structure diagram of the improved optimization recognition model is as follows: Figure 2As shown in the figure, Conv represents the convolution layer for feature extraction; C2f represents the feature fusion module based on CSPNet, which is used for more efficient feature transfer; ASFF_2 represents a double-layer adaptive feature fusion layer, which adaptively fuses multi-scale features by learning weights to solve the inconsistency problem between features of different scales; ASFF_3 represents a three-layer adaptive feature fusion layer, which adaptively fuses multi-scale features by learning weights to solve the inconsistency problem between features of different scales; Basic Block represents the basic building block, which is a module combining convolution, normalization, activation function and CSP structure; Detect represents the detection head, which is responsible for the final target bounding box and category prediction, and SPPF is the abbreviation of Spatial Pyramid Pooling Fast, which is an improved pooling module, mainly used to process multi-scale feature fusion and optimize computational efficiency.

[0082] In this embodiment, in order to effectively evaluate the performance of the improved model, the following evaluation indicators are used: accuracy (Precision), recall (Recall), mean average precision (mAP), model parameter quantity (Params) and total floating-point operations (FLOPs). Accuracy is used to measure the proportion of all instances predicted as positive samples that are actually positive samples. Recall is used to evaluate the model's ability to successfully predict in actual positive samples. Mean average precision (mAP) represents the average of the AP values ​​of each category and is a comprehensive indicator for evaluating the overall performance of the algorithm in different categories. Among them, the model parameter quantity refers to the total number of all trainable parameters in the neural network, including weights and biases. These parameters will be updated during the training process for model optimization and prediction. The larger the number of parameters, the more complex the model is usually and has stronger expressive power. The total floating-point operations are one of the criteria for measuring the computational complexity of the model. The larger the FLOPs, the more computing resources are required for model reasoning and training. Table 3 is a comparison result table of the YOLOv8 original model and the YOLOv8 optimized recognition model in this embodiment. The comparison result of the YOLOv8 original model and the YOLOv8 optimized recognition model in this embodiment is shown in Table 3:

[0083] Table 3

[0084]

[0085] By comparing the original YOLOv8 model with the improved optimized recognition model, it can be clearly seen that the optimized recognition model shows significant advantages in all indicators. The accuracy of the optimized recognition model increased by 39.3%, the recall rate increased by 7.6%, and the mean average precision (mAP) increased by 11.5%. These results show that the prediction accuracy and comprehensive detection performance of the improved optimized recognition model in the target detection task have been greatly improved. At the same time, the number of model parameters has been reduced by 78.9%, and the total floating-point operations have been reduced by 82.3%, which greatly reduces the complexity and computational cost of the model. The above improvements not only effectively improve the performance of the model, but also enhance its application potential on resource-constrained devices, such as mobile devices and embedded systems. Overall, the improved optimized recognition model achieves lightweight and high efficiency while maintaining high accuracy, and has a wider range of practical application prospects.

[0086] In this embodiment, Figure 3 is the image to be detected, Figure 4 This is the image detected by the original YOLOv8 model. Figure 5 To optimize the image after the recognition model is detected. Figures 3 to 5 It can be found that the detection accuracy of the original YOLOv8 model in the presence of noise and small targets is relatively weaker than that of the improved YOLOv8 model, and the detection accuracy of the image detected by the optimized recognition model is higher.

[0087] Preferably, the optimization loss function is a SIoU function, and the SIoU function is also used to dynamically adjust the penalty term of the center point distance. The penalty term calculation formula of the center point distance is configured as:

[0088] ,

[0089] in, The penalty term used to represent the distance from the center point, and They are used to represent the central horizontal coordinate and central vertical coordinate of the prediction box of the target detection task, and They are used to represent the central horizontal coordinate and central vertical coordinate of the real box of the target detection task respectively.

[0090] Preferably, the penalty calculation formula for scale and aspect ratio is configured as:

[0091] ,

[0092] in, A penalty term for scale and aspect ratio, and They are used to represent the width and height of the prediction box respectively. and They are used to represent the width and height of the real frame respectively;

[0093] The penalty calculation formula for the angle is configured as:

[0094] ,

[0095] in, A penalty term for angles, and They are used to represent the angles between the predicted box and the true box;

[0096] The formula configuration of SIoU function is:

[0097] ,

[0098] in, Used to represent the SIoU function, Used to represent the traditional IoU function, , and They are respectively used to represent the preset first weight coefficient, second weight coefficient and third weight coefficient.

[0099] Specifically, in this embodiment, by incorporating the SIoU function, the YOLOv8 model can fit the target boundary more accurately when processing irregular or non-horizontal targets. Secondly, the SIoU function solves the problem that the traditional IoU function is not sensitive enough when detecting small targets by dynamically adjusting the target frame scale. Experiments show that the SIoU function improves the regression accuracy of the YOLOv8 model for small targets and targets with large aspect ratio differences, while reducing the mismatch between the frame and the target size, thereby effectively reducing the missed detection rate and the false detection rate.

[0100] In summary, the introduction of the SIoU function makes the target positioning of the YOLOv8 model more accurate, especially in complex scenes. According to the experimental results in Table 1, after integrating the SIoU function, the mean average precision (mAP) of the YOLOv8 model on the pipeline shallow section image dataset is further improved, and it has better generalization ability when dealing with targets of different scales and shapes.

[0101] Embodiment 2 is the second embodiment of the present invention. Different from the previous embodiment, this embodiment provides an image processing module 6, which can further improve the detection accuracy and robustness of the model. It includes an image processing module 6 connected to a data acquisition module 3, which is used to perform background separation, image denoising and target enhancement on the original image data in the training set, test set and validation set in turn to obtain optimized image data.

[0102] Working principle of embodiment 2:

[0103] The specific processing steps of the image processing module 6 are as follows:

[0104] Background separation:

[0105] Due to the complex seabed environment, background interference (such as marine sediments, seaweed, rocks, etc.) may affect the accuracy of pipeline detection, so background separation is required to highlight the target area.

[0106] Method: Gaussian mixture model is used for background modeling to separate pipeline targets from background.

[0107] Image Denoising:

[0108] Since seabed images are affected by water media and are often accompanied by noise (such as scattering noise, uneven illumination noise, etc.), denoising is required to improve image clarity.

[0109] Method: Adaptive wavelet denoising: decompose the image at multiple scales, remove high-frequency noise and improve image quality.

[0110] Target Enhancement:

[0111] Since targets in seabed images may have problems such as low contrast and blurred edges, target enhancement is required to make key targets such as pipelines clearer.

[0112] Method: Adaptive histogram equalization: avoids the problem of excessive noise enhancement in ordinary histogram equalization and makes the target details clearer.

[0113] Embodiment 3 is the third embodiment of the present invention. Different from the previous embodiment, this embodiment provides an environment detection module 7, which can improve the adaptability of the target detection model in a complex seabed environment, while improving the detection accuracy and reducing the false detection rate. It includes an environment detection module 7 connected to the training module 4. The environment detection module 7 is used to detect the light brightness and light uniformity of the seabed environment where the pipeline is located, and detect the relative distance from the pipeline;

[0114] The training module 4 inputs the light illuminance, light uniformity, and relative distance into a preset learning rate optimization formula to calculate an optimized learning rate, and inputs the light illuminance, light uniformity, and relative distance into a preset weight optimization formula to calculate an optimized weight attenuation parameter;

[0115] The optimized recognition model is output after improving the model structure by optimizing the learning rate and optimizing the weight decay parameters.

[0116] Working principle of embodiment 3:

[0117] The detection of light brightness is achieved by calculating the average grayscale value of the pixels in the image, and the detection of light uniformity is achieved by calculating the local window mean difference. The smaller the local window mean difference, the higher the light uniformity. The relative distance is detected by the depth camera. The optimized learning rate and optimized weight decay parameters are calculated by light brightness, light uniformity, and relative distance to further improve the model structure of the optimized recognition model, filtering out the influence of complex changes in lighting and uneven target distance on the model recognition accuracy, so that the improved optimized recognition model can still maintain high accuracy in the case of complex changes in lighting and uneven target distance, thereby improving the adaptability of the target detection model in complex seabed environments.

[0118] Preferably, the learning rate optimization formula is configured as:

[0119] ,

[0120] in, Used to represent the optimized learning rate, Used to represent the initial learning rate of the optimized recognition model, It is used to indicate the adjustment coefficient of light intensity on learning rate, reflecting the influence of light on learning rate optimization; Used to indicate light brightness. An exponential adjustment factor used to represent the effect of light intensity on the learning rate, controlling the intensity of the effect of light on the learning rate; It is used to indicate the relative distance influence coefficient, and adjusts the relationship between illumination uniformity and pipeline relative distance to influence the learning rate. Used to indicate the uniformity of lighting. Used to indicate relative distance. Used to represent the distance factor adjustment coefficient, which controls the influence of relative distance on the learning rate.

[0121] Specifically, the key parameters in this embodiment are adjusted as follows:

[0122] Light intensity: In low-light environments (such as deep under the sea), the light intensity is small, which may lead to low learning efficiency during training. Therefore, the optimization formula and It needs to be adjusted so that the learning rate increases in low-light environments to speed up the convergence of the model.

[0123] Lighting uniformity: Lighting uniformity directly affects the clarity of pipeline images. If the lighting distribution is uneven ( ≈0), the recognition effect of the model will be affected, so the relative distance affects the coefficient The learning rate is optimized to enhance the model's learning ability under uneven lighting conditions.

[0124] Relative distance to pipeline: When the pipeline is far away from the camera, the clarity and details of the image may be incomplete, affecting the recognition effect. In this case, a larger value of the relative distance will reduce the learning rate, while a smaller relative distance value (closer pipeline) will increase the learning rate to better capture image details.

[0125] The parameters of one of the images in a set of submarine pipeline shallow profile image data are as follows:

[0126] Light brightness: =0.4

[0127] Light uniformity: =0.8

[0128] Relative distance from pipeline: =5.0

[0129] Substituting these values ​​into the formula, we set the initial learning rate to =0.001, adjustment coefficient of light intensity =0.5, light brightness index influence coefficient =1.2, Light uniformity influence factor =0.3, relative distance adjustment coefficient =0.7;

[0130] By calculation, the optimized learning rate is obtained , this learning rate will be used for the next update in the training process.

[0131] Implementation effect:

[0132] Improved training results: By dynamically adjusting the learning rate, the model can better adapt to different environmental lighting conditions and pipeline locations, improving the accuracy of the YOLOv8 model in submarine pipeline identification.

[0133] Convergence speed: The optimized learning rate can speed up the convergence of the model, especially in low-light or uneven-light environments, the model can be trained more stably.

[0134] Preferably, the weight optimization formula is configured as:

[0135] ,

[0136] in, Used to represent the optimized weight decay coefficient, It is used to represent the initial weight attenuation coefficient of the optimized recognition model and set the initial value of the attenuation parameter; It is used to represent the adjustment coefficient of light brightness on weight attenuation, and adjust the intensity of the influence of light on the attenuation parameter; It is used to indicate the influence index of light brightness on attenuation parameters and adjust the influence intensity of light on weight attenuation. The adjustment coefficient used to represent the uniformity of illumination, indicating the intensity of the influence of the uniformity of illumination on the weight attenuation; It is used to indicate the attenuation parameter adjustment coefficient related to pipeline distance, and controls the influence of relative position on attenuation; Used to represent the preset time decay coefficient and control the speed of the exponential decay function; Used to represent time variables, An adjustment factor used to represent the effect of time factors on attenuation.

[0137] Specifically, in this embodiment, the light brightness affects the quality of the image and indirectly affects the stability and effect of model training. When the light brightness is high, the optimized weight attenuation coefficient increases, thereby improving the regularization strength of the model and preventing overfitting; when the light brightness is low, the attenuation coefficient decreases appropriately, allowing the model to learn the characteristics of darker environments in more detail.

[0138] The influence of illumination uniformity is reflected in the clarity of the image. In an environment with uniform illumination, the image quality is good, and the optimized weight decay coefficient will be small, which helps the model to learn features in a refined manner; while in an environment with uneven illumination, the weight decay will increase to reduce the possible overfitting of the model.

[0139] The relative position of the pipeline and the camera affects the recognition difficulty of the model. When the pipeline is far away, the pipeline features in the image will become blurred, and the model needs stronger regularization to prevent overfitting. Therefore, the optimized attenuation coefficient will increase, enhancing the regularization effect of the model.

[0140] Time decay : As training progresses, the exponential decay function controls the gradual decrease of weight decay. In the early stage of training, a larger weight decay can prevent overfitting; in the later stage of training, the decay gradually decreases, allowing the model to fit the data more flexibly. To further enhance the control of time-dependent attenuation and make training more efficient.

[0141] The value range of is [0,∞), which represents the optimized weight decay value and affects the strength of model regularization. A larger optimized weight decay coefficient value indicates that the model regularization is stronger and prevents overfitting. A smaller optimized weight decay coefficient value helps improve the model's fitting ability, but may bring the risk of overfitting.

[0142] Practical Application:

[0143] Optimization of lighting conditions: In a strong lighting environment, the effect of lighting on weight attenuation increases, so the attenuation parameter will be increased appropriately to enhance regularization and prevent overfitting. In a weak lighting environment, the model requires a higher attenuation parameter to prevent overfitting in low-light conditions.

[0144] The influence of illumination uniformity: Under uniform illumination conditions, the model can stably identify features, so the attenuation value is small. Under uneven illumination conditions, the attenuation value needs to be increased to enhance the generalization ability of the model.

[0145] Influence of pipeline position: As the pipeline distance increases, the optimized attenuation value will increase to prevent overfitting of the training results by the long-distance pipeline features.

[0146] Control of time decay: As the training process progresses, the time decay factor enables the optimization process to gradually reduce weight decay, thereby adapting to more detailed feature learning and preventing overfitting.

[0147] A pipeline shallow section image automatic recognition method based on deep learning is applied to the above-mentioned pipeline shallow section image automatic recognition system based on deep learning, such as Figure 6 As shown, including:

[0148] Step S1, introducing module 1 to introduce the YOLOv8 model as the basic model;

[0149] Step S2, the model improvement module 2 integrates the attention feature pyramid network, the target detection dynamic head structure and the optimization loss function into the basic model to form an initial recognition model, the attention feature pyramid network dynamically adjusts the weight of feature extraction according to different feature maps, the target detection dynamic head structure automatically adjusts the parameter configuration of the detection head according to the feature complexity of the input image, and the optimization loss function dynamically adjusts the penalty items of angle, scale and aspect ratio to optimize the direction error of the detection frame;

[0150] Step S3, the data acquisition module 3 acquires a plurality of submarine pipeline shallow profile images, and annotates them to form a pipeline shallow profile image dataset, and then divides the pipeline shallow profile image dataset into a training set, a test set, and a verification set;

[0151] Step S4, the training module 4 trains the initial recognition model according to the data in the training set, and adjusts the hyperparameters in the initial recognition model training process according to the data in the validation set, until the performance parameters of the model meet the preset parameter conditions, and outputs it as an optimized recognition model;

[0152] Step S5, the recognition module 5 recognizes the image data in the test set according to the optimized recognition model to obtain the pipeline shallow section image recognition result.

[0153] The above are only preferred embodiments of the present invention. The protection scope of the present invention is not limited to the above embodiments. All technical solutions under the concept of the present invention belong to the protection scope of the present invention. It should be pointed out that for ordinary technicians in this technical field, some improvements and modifications without departing from the principle of the present invention should also be regarded as the protection scope of the present invention.

Claims

1. A pipeline shallow section image automatic recognition system based on deep learning, characterized in that: include: An introduction module (1) is used to introduce the YOLOv8 model as a basic model; A model improvement module (2), connected to the introduction module (1), is used to integrate the attention feature pyramid network, the target detection dynamic head structure and the optimization loss function into the basic model to form an initial recognition model; The attention feature pyramid network dynamically adjusts the weight of feature extraction according to different feature maps, the target detection dynamic head structure automatically adjusts the parameter configuration of the detection head according to the feature complexity of the input image, and the optimization loss function dynamically adjusts the penalty terms of angle, scale and aspect ratio to optimize the direction error of the detection frame; A data acquisition module (3) is used to acquire a plurality of submarine pipeline shallow section images, annotate them to form a pipeline shallow section image data set, and further divide the pipeline shallow section image data set into a training set, a test set and a verification set; A training module (4), connected to the data acquisition module (3) and the model improvement module (2), respectively, for training the initial recognition model according to the data in the training set, and adjusting hyperparameters in the initial recognition model training process according to the data in the validation set, until the performance parameters of the model meet the preset parameter conditions and outputting the model as an optimized recognition model; The recognition module (5) is connected to the training module (4) and the data acquisition module (3) and is used to recognize the image data in the test set according to the optimized recognition model to obtain pipeline shallow section image recognition results.

2. According to the deep learning-based pipeline shallow section image automatic recognition system of claim 1, it is characterized by: The optimization loss function is a SIoU function, and the SIoU function is also used to dynamically adjust the penalty term of the center point distance. The penalty term calculation formula of the center point distance is configured as: , in, A penalty term for the distance from the center point, and They are used to represent the central horizontal coordinate and central vertical coordinate of the prediction box of the target detection task, and They are used to represent the central horizontal coordinate and central vertical coordinate of the real box of the target detection task respectively.

3. The pipeline shallow section image automatic recognition system based on deep learning according to claim 2 is characterized by: The penalty calculation formula for the scale and aspect ratio is configured as: ； in, A penalty term for the scale and aspect ratio, and are used to represent the width and height of the prediction box respectively, and are used to represent the width and height of the real frame respectively; The penalty calculation formula for the angle is configured as: , in, A penalty term for the angle, and are respectively used to represent the angle between the predicted frame and the real frame; The formula configuration of the SIoU function is: ， in, Used to represent the SIoU function, Used to represent the traditional IoU function, , and They are respectively used to represent the preset first weight coefficient, second weight coefficient and third weight coefficient.

4. The pipeline shallow section image automatic recognition system based on deep learning according to claim 1 is characterized by: The ratio of data in the training set, the test set and the validation set is 10:5:

1.

5. The pipeline shallow section image automatic recognition system based on deep learning according to claim 1 is characterized by: The performance parameters include accuracy, recall, average precision, model parameter quantity and total floating point operation quantity.

6. The pipeline shallow section image automatic recognition system based on deep learning according to claim 1 is characterized in that: It also includes an image processing module (6) connected to the data acquisition module (3) and used to perform background separation, image denoising and target enhancement on the original image data in the training set, the test set and the verification set in sequence, so as to obtain optimized image data.

7. The pipeline shallow section image automatic recognition system based on deep learning according to claim 1 is characterized in that: It also includes an environment detection module (7) connected to the training module (4), and the environment detection module (7) is used to detect the light brightness and light uniformity of the seabed environment where the pipeline is located, and detect the relative distance from the pipeline; The training module (4) inputs the light luminance, the light uniformity, and the relative distance into a preset learning rate optimization formula to calculate an optimized learning rate, and inputs the light luminance, the light uniformity, and the relative distance into a preset weight optimization formula to calculate an optimized weight attenuation parameter; The optimized recognition model improves the model structure according to the optimized learning rate and the optimized weight decay parameter and then outputs it.

8. The pipeline shallow section image automatic recognition system based on deep learning according to claim 7 is characterized by: The learning rate optimization formula is configured as: , in, is used to denote the optimization learning rate, It is used to represent the initial learning rate of the optimized recognition model, It is used to represent the adjustment coefficient of the light brightness to the learning rate, Used to indicate the brightness of the light. An exponential adjustment factor for representing the effect of the light intensity on the learning rate, Used to express the relative distance influence coefficient, Used to indicate the uniformity of illumination, Used to indicate the relative distance, Used to represent the distance factor adjustment coefficient.

9. The pipeline shallow section image automatic recognition system based on deep learning according to claim 8 is characterized by: The weight optimization formula is configured as: ,in, Used to represent the optimization weight decay coefficient, is used to represent the initial weight decay coefficient of the optimized recognition model, It is used to represent the adjustment coefficient of the light brightness to the weight attenuation. It is used to indicate the influence index of the light brightness on the attenuation parameter. An adjustment factor used to indicate the uniformity of illumination. It is used to represent the attenuation parameter adjustment coefficient related to the pipeline distance. Used to indicate the preset time decay coefficient, Used to represent time variables, An adjustment factor used to represent the effect of time factors on attenuation.

10. A pipeline shallow section image automatic recognition method based on deep learning, applied to the pipeline shallow section image automatic recognition system based on deep learning according to any one of claims 1 to 9, characterized in that: include: Step S1, introducing module (1) to introduce the YOLOv8 model as the basic model; Step S2, the model improvement module (2) integrates the attention feature pyramid network, the target detection dynamic head structure and the optimization loss function into the basic model to form an initial recognition model, the attention feature pyramid network dynamically adjusts the weight of feature extraction according to different feature maps, the target detection dynamic head structure automatically adjusts the parameter configuration of the detection head according to the feature complexity of the input image, and the optimization loss function dynamically adjusts the penalty items of angle, scale and aspect ratio to optimize the direction error of the detection frame; Step S3, the data acquisition module (3) acquires a plurality of submarine pipeline shallow profile images, and annotates them to form a pipeline shallow profile image data set, and then divides the pipeline shallow profile image data set into a training set, a test set, and a verification set; Step S4, the training module (4) trains the initial recognition model according to the data in the training set, and adjusts the hyperparameters in the initial recognition model training process according to the data in the validation set, until the performance parameters of the model meet the preset parameter conditions and outputs it as an optimized recognition model; Step S5, the recognition module (5) recognizes the image data in the test set according to the optimized recognition model to obtain a pipeline shallow section image recognition result.

Images