Convolutional Attention Research Articles

A customised ConvNeXt-SCC network: integrating improved principal component analysis with ConvNeXt to enhance tire crown defect detection

ABSTRACT Defects such as bubbles and rubber shortages occasionally occur during vehicle tire production. Tiny defects in the crown gradually expand under load, ultimately causing tire failure. However, machine-vision-based crown defect detection faces challenges like limited datasets, varying groove patterns, low colour contrast, and small defect sizes. Therefore, this article integrates principal component analysis (PCA) and deep learning techniques for the visual detection of tire crown defects. A dataset of tire crown defects suitable for deep learning was constructed using an improved-PCA method and composite sampling. We propose the ConvNeXt-Shallow Conv CBAM (ConvNeXt-SCC) model, which includes an improved convolutional block attention module (CBAM) in a shallow stacked structure. And we propose an efficient and accurate method called Improved Selective Search (ISS) to detect and locate tire crown defective regions. The ISS pre-selected box generation method, enhanced with a priori knowledge filtering module, was used alongside a classification model to achieve online inference detection of large-size tire crown images. The experimental results show that the performance indices of the proposed ConvNeXt-Shallow Conv CBAM (ConvNeXt-SCC) model significantly improve compared to the original ConvNeXt-T. The defect detection and localisation method based on ISS meets the online inspection requirements of tire production.

An improved algorithm for salient object detection of microscope based on U2-Net.

With the rapid advancement of modern medical technology, microscopy imaging systems have become one of the key technologies in medical image analysis. However, manual use of microscopes presents issues such as operator dependency, inefficiency, and time consumption. To enhance the efficiency and accuracy of medical image capture and reduce the burden of subsequent quantitative analysis, this paper proposes an improved microscope salient object detection algorithm based on U2-Net, incorporating deep learning technology. The improved algorithm first enhances the network's key information extraction capability by incorporating the Convolutional Block Attention Module (CBAM) into U2-Net. It then optimizes network complexity by constructing a Simple Pyramid Pooling Module (SPPM) and uses Ghost convolution to achieve model lightweighting. Additionally, data augmentation is applied to the slides to improve the algorithm's robustness and generalization. The experimental results show that the size of the improved algorithm model is 72.5MB, which represents a 56.85% reduction compared to the original U2-Net model size of 168.0MB. Additionally, the model's prediction accuracy has increased from 92.24 to 97.13%, providing an efficient means for subsequent image processing and analysis tasks in microscopy imaging systems.

Remaining useful life prognostics of bearings based on convolution attention networks and enhanced transformer

Rolling bearings are critical components of industrial equipment, and predicting their remaining useful life (RUL) is a challenging task. This article proposes a new end-to-end model (MDSCT) to achieve efficient and accurate prediction of bearing RUL. MDSCT uses the raw vibration signals collected by sensors for prediction, without relying on a large amount of prior knowledge. The feature extraction backbone of this model is composed of an MDSC attention module that integrates multiple parallel depth-wise separable convolutions and an efficient attention mechanism, combined with a tansformer encoder (PPSformer) optimized using patch embedding and probesparse self attention techniques. They are respectively used to capture local subtle features and global dependent features in degraded signals. This article also proposes an improved adaptive activation function AdaptH_Swish to enhance the model's ability to model nonlinear relationships. To comprehensively verify the comprehensive performance of the model, this paper conducted detailed ablation and comparative experiments using two standard datasets, PHM2012 and XJTU-SY. The experimental results not only confirm the rationality and efficiency of the model structure design, but also demonstrate significant advantages in three evaluation indicators compared to various existing methods, fully demonstrating the high generalization and strong robustness of the MDSCT model in bearing RUL prediction tasks.

TW-YOLO: An Innovative Blood Cell Detection Model Based on Multi-Scale Feature Fusion.

As deep learning technology has progressed, automated medical image analysis is becoming ever more crucial in clinical diagnosis. However, due to the diversity and complexity of blood cell images, traditional models still exhibit deficiencies in blood cell detection. To address blood cell detection, we developed the TW-YOLO approach, leveraging multi-scale feature fusion techniques. Firstly, traditional CNN (Convolutional Neural Network) convolution has poor recognition capabilities for certain blood cell features, so the RFAConv (Receptive Field Attention Convolution) module was incorporated into the backbone of the model to enhance its capacity to extract geometric characteristics from blood cells. At the same time, utilizing the feature pyramid architecture of YOLO (You Only Look Once), we enhanced the fusion of features at different scales by incorporating the CBAM (Convolutional Block Attention Module) in the detection head and the EMA (Efficient Multi-Scale Attention) module in the neck, thereby improving the recognition ability of blood cells. Additionally, to meet the specific needs of blood cell detection, we designed the PGI-Ghost (Programmable Gradient Information-Ghost) strategy to finely describe the gradient flow throughout the process of extracting features, further improving the model's effectiveness. Experiments on blood cell detection datasets such as BloodCell-Detection-Dataset (BCD) reveal that TW-YOLO outperforms other models by 2%, demonstrating excellent performance in the task of blood cell detection. In addition to advancing blood cell image analysis research, this work offers strong technical support for future automated medical diagnostics.

Motor Fault Diagnosis Based on Convolutional Block Attention Module-Xception Lightweight Neural Network.

Electric motors play a crucial role in self-driving vehicles. Therefore, fault diagnosis in motors is important for ensuring the safety and reliability of vehicles. In order to improve fault detection performance, this paper proposes a motor fault diagnosis method based on vibration signals. Firstly, the vibration signals of each operating state of the motor at different frequencies are measured with vibration sensors. Secondly, the characteristic of Gram image coding is used to realize the coding of time domain information, and the one-dimensional vibration signals are transformed into grayscale diagrams to highlight their features. Finally, the lightweight neural network Xception is chosen as the main tool, and the attention mechanism Convolutional Block Attention Module (CBAM) is introduced into the model to enforce the importance of the characteristic information of the motor faults and realize their accurate identification. Xception is a type of convolutional neural network; its lightweight design maintains excellent performance while significantly reducing the model's order of magnitude. Without affecting the computational complexity and accuracy of the network, the CBAM attention mechanism is added, and Gram's corner field is combined with the improved lightweight neural network. The experimental results show that this model achieves a better recognition effect and faster iteration speed compared with the traditional Convolutional Neural Network (CNN), ResNet, and Xception networks.

Temperature compensation method for impedance signals of bolt loosening under temperature varying conditions using TransUnet

Abstract Piezoelectric transducer based electro-mechanical impedance technology is an effective and reliable detection method for bolt structure loosening under external load. However, temperature change usually causes some non-loosening factors to change the impedance spectrum of bolt structure. In order to reduce the misjudgment of loosening damage caused by temperature change, it is necessary to construct a temperature compensation model for impedance spectrum. In this paper, the convolutional structure in U-net and Transformer are effectively combined to form a TransUnet deep neural network structure suitable for input and output of one-dimensional data. Using impedance data between 10 °C and 50 °C and temperature as input to the network. After convolution operation, the convolutional block attention module is embedded in the U-net to optimize the encoder transmission characteristics and enhance the performance of the skip connection in the traditional U-shaped structure. The temperature compensation rate (TCR) is defined to measure the effect of temperature compensation. Then the lightweight convolutional neural network structure is used to recognize the bolt loosening damage of the compensated impedance signal. The generalization ability of the TransUnet was tested using impedance data that is not in the training dataset. The results show that the TransUnet proposed in the paper can realize the temperature compensation of multi-peak impedance signals. The TCR and recognition accuracy of bolt loose damage reaches reach 0.003 and 95.7%, respectively, which is 11% higher than that without temperature compensation. At 60 °C, the TCR and the identification accuracy of loosening damage can still reach 0.0055 and 90.2%, respectively, which show that the TransUnet has strong generalization ability.

ECM-YOLO: a real-time detection method of steel surface defects based on multiscale convolution

Steel surface defects, characterized by multiple types, varied scales, and overlapping occurrences, directly impact the quality, performance, and reliability of industrial products. Proposing a high-precision and high-speed steel surface defect detection algorithm is crucial for ensuring product quality. In this regard, this paper introduces ECM-YOLO, a detection network based on YOLOv8n. First, addressing the insufficient information capture of the C2f module, the C2f enhanced multiscale convolution processing (C2f_EMSCP) module is proposed, enhancing global and local feature capture capabilities through multiscale convolutions. Second, to further enhance the network’s robustness and focus on critical information, the channel prior convolutional attention (CPCA) mechanism is integrated between the backbone and neck networks to facilitate more efficient information transmission. Last, a novel, to the best of our knowledge, detection head, i.e., multiscale simple and efficient anchor matching head (MultiSEAMHead), is proposed to mitigate accuracy issues arising from overlaps between different types of defects. Experimental results demonstrate that ECM-YOLO achieves mAPs of 78.9% and 68.2% on the NEU-DET and GC 10-DET data sets, respectively, outperforming YOLOv8n by 2.5% and 4.4%. Moreover, ECM-YOLO excels in model parameters, computational efficiency, and inference speed compared with other models. These findings highlight the applicability of ECM-YOLO for real-time defect detection in industrial settings.

SCS-YOLO: A Defect Detection Model for Cigarette Appearance

Appearance defects significantly impact cigarette quality. However, in the current high-speed production lines, manual inspection and traditional methods are unable to satisfy the actual demands of inspection. Therefore, a real-time and high-precision defect detection model for cigarette appearance, SCS-YOLO, is presented. The model integrates space-to-depth convolution (SPD-Conv), a convolutional block attention module (CBAM), and a self-calibrated convolutional module (SCConv). SPD-Conv replaces the pooling structure to enhance the granularity of feature information. CBAM improves the ability to pay attention to defect locations. Improved self-calibrated convolution broadens the network’s receptive field and feature fusion capability. Additionally, Complete IoU loss (CIoU) is replaced with Efficient IoU Loss (EIoU) to enhance model localization and mitigate sample imbalance. The experimental results show that the accuracy of SCS-YOLO is 95.5% and the mAP (mean average precision) value is 95.2%. Compared with the original model, the accuracy and mAP value of the SCS-YOLO model are improved by 4.0%. Furthermore, the model achieves a detection speed of 216 FPS, meeting cigarette production lines’ accuracy and speed demands. Our research will positively impact the real-time detection of appearance defects in cigarette production lines.

A Robust Generalized Zero-Shot Learning Method with Attribute Prototype and Discriminative Attention Mechanism

In the field of Generalized Zero-Shot Learning (GZSL), the challenge lies in learning attribute-based information from seen classes and effectively conveying this knowledge to recognize both seen and unseen categories during the training process. This paper proposes an innovative approach to enhance the generalization ability and efficiency of GZSL models by integrating a Convolutional Block Attention Module (CBAM). The CBAM blends channel-wise and spatial-wise information to emphasize key features, thereby improving the model’s discriminative and localization capabilities. Additionally, the method employs a ResNet101 backbone for systematic image feature extraction, enhanced contrastive learning, and a similarity map generator with attribute prototypes. This comprehensive framework aims to achieve robust visual–semantic embedding for classification tasks. The proposed method demonstrates significant improvements in performance metrics in benchmark datasets, showcasing its potential in advancing GZSL applications.

Detection of small foreign objects in Pu-erh sun-dried green tea: An enhanced YOLOv8 neural network model based on deep learning

To efficiently and accurately detect minuscule foreign objects in the processing of Pu-erh sun-dried green tea, ensuring food quality and consumer safety, this study innovatively proposes an enhanced YOLOv8 neural network based on deep learning. In light of the shortcomings of the traditional YOLOv8 network, to further enhance the model's ability to identify foreign object targets, improve the depth and breadth of feature extraction, and to better understand the contextual connections between different parts of the image, this study employs a Shape-IoU optimized loss function. It replaces parts of the network structure with Receptive-Field Attention Convolution technology and embeds Double Attention Networks to optimize the network. Experimental results show that the enhanced YOLOv8 neural network model achieves a precision rate of 98.35% for the detection of foreign objects in Pu-erh sun-dried green tea, which is a 3.93% increase compared to the original YOLOv8 network. Compared to mainstream detection models such as YOLOv7, YOLOv5, Faster-RCNN, CornerNet, and SSD, the mean Average Precision values of the enhanced YOLOv8 network model have significantly increased by 4.48%, 6.66%, 13.63%, 13.20%, and 9.84% respectively. This enhanced YOLOv8 network provides a viable research method and significant reference for the detection of small foreign objects in Pu-erh sun-dried green tea, holding substantial importance for tea-producing enterprises and food safety regulatory authorities. Furthermore, this study offers a more comprehensive and efficient solution for foreign object detection in Pu-erh tea and the broader food industry, laying the groundwork for the modernization and intelligentization of food safety and quality control.

Research and Experiment on a Chickweed Identification Model Based on Improved YOLOv5s

Currently, multi-layer deep convolutional networks are mostly used for field weed recognition to extract and identify target features. However, in practical application scenarios, they still face challenges such as insufficient recognition accuracy, a large number of model parameters, and slow detection speed. In response to the above problems, using chickweed as the identification object, a weed identification model based on improved YOLOv5s was proposed. Firstly, the Squeeze-and-Excitation Module (SE) and Convolutional Block Attention Module (CBAM) were added to the model’s feature extraction network to improve the model’s recognition accuracy; secondly, the Ghost convolution lightweight feature fusion network was introduced to effectively identify the volume, parameter amount, and calculation amount of the model, and make the model lightweight; finally, we replaced the loss function in the original target bounding box with the Efficient Intersection over Union (EloU) loss function to further improve the detection performance of the improved YOLOv5s model. After testing, the accuracy of the improved YOLOv5s model was 96.80%, the recall rate was 94.00%, the average precision was 93.20%, and the frame rate was 14.01 fps, which were improved by 6.6%, 4.4%, 1.0%, and 6.1%, respectively, compared to the original YOLOv5s model. The model volume was 9.6 MB, the calculation amount was 13.6 GB, and the parameter amount was 5.9 MB, which decreased by 29.4%, 14.5%, and 13.2% compared with the original YOLOv5s model, respectively. This model can effectively distinguish chickweed between crops. This research can provide theoretical and technical support for efficient identification of weeds in complex field environments.

Method for Augmenting Side-Scan Sonar Seafloor Sediment Image Dataset Based on BCEL1-CBAM-INGAN.

In this paper, a method for augmenting samples of side-scan sonar seafloor sediment images based on CBAM-BCEL1-INGAN is proposed, aiming to address the difficulties in acquiring and labeling datasets, as well as the insufficient diversity and quantity of data samples. Firstly, a Convolutional Block Attention Module (CBAM) is integrated into the residual blocks of the INGAN generator to enhance the learning of specific attributes and improve the quality of the generated images. Secondly, a BCEL1 loss function (combining binary cross-entropy and L1 loss functions) is introduced into the discriminator, enabling it to focus on both global image consistency and finer distinctions for better generation results. Finally, augmented samples are input into an AlexNet classifier to verify their authenticity. Experimental results demonstrate the excellent performance of the method in generating images of coarse sand, gravel, and bedrock, as evidenced by significant improvements in the Frechet Inception Distance (FID) and Inception Score (IS). The introduction of the CBAM and BCEL1 loss function notably enhances the quality and details of the generated images. Moreover, classification experiments using the AlexNet classifier show an increase in the recognition rate from 90.5% using only INGAN-generated images of bedrock to 97.3% using images augmented using our method, marking a 6.8% improvement. Additionally, the classification accuracy of bedrock-type matrices is improved by 5.2% when images enhanced using the method presented in this paper are added to the training set, which is 2.7% higher than that of the simple method amplification. This validates the effectiveness of our method in the task of generating seafloor sediment images, partially alleviating the scarcity of side-scan sonar seafloor sediment image data.

Fundus-DANet: Dilated Convolution and Fusion Attention Mechanism for Multilabel Retinal Fundus Image Classification

The difficulty of classifying retinal fundus images with one or more illnesses present or missing is known as fundus multi-lesion classification. The challenges faced by current approaches include the inability to extract comparable morphological features from images of different lesions and the inability to resolve the issue of the same lesion, which presents significant feature variances due to grading disparities. This paper proposes a multi-disease recognition network model, Fundus-DANet, based on the dilated convolution. It has two sub-modules to address the aforementioned issues: the interclass learning module (ILM) and the dilated-convolution convolutional block attention module (DA-CBAM). The DA-CBAM uses a convolutional block attention module (CBAM) and dilated convolution to extract and merge multiscale information from images. The ILM uses the channel attention mechanism to map the features to lower dimensions, facilitating exploring latent relationships between various categories. The results demonstrate that this model outperforms previous models in classifying fundus multilocular lesions in the OIA-ODIR dataset with 93% accuracy.

A Multi-Mechanism Fusion Method for Robust State of Charge Estimation via Bidirectional Long Short-Term Memory Model with Mixture Kernel Mean p-Power Error Loss Optimized by Golden Jackal Optimization Algorithm

Abstract Accurate state of charge (SOC) estimation is crucial for effective battery management in various applications. The bidirectional long short-term memory (BiLSTM) as an outstanding nonlinear regression model can be used for SOC estimation. This work develops a novel multi-mechanism fusion method based on BiLSTM to further enhance its estimation performance for SOC, in which the convolutional neural network (CNN), attention mechanism, and mixture kernel mean p-power error (MKMPE) loss are introduced into the BiLSTM framework for addressing different issues. First, the introduction of CNN components aims to extract essential features from battery data, enhancing the model's comprehension of complex information. Then, the attention mechanism is used to further refine the model's perceptual ability and a robust MKMPE loss is introduced into the BiLSTM framework to replace its original mean squared error loss, and a novel robust model is developed to suppress non-Gaussian noise interference. Finally, some key hyperparameters of the proposed model are fine-tuned using the golden jackal optimization algorithm, resulting in improved estimation performance. Comparative numerical experiments are meticulously conducted in various cases to evaluate the performance of the proposed method, and the experiment results demonstrate that it can perform outstanding effectiveness in handling non-Gaussian noise scenarios.

Dataset-level color augmentation and multi-scale exploration methods for polyp segmentation

Automatic segmentation of polyps from colonoscopy images plays a critical role in early screening and treatment of colorectal cancer. Although deep learning methods have made significant progress, precise polyp segmentation faces two challenges: (1) the imbalance of color appearances in the limited training dataset hinders the generalization of the model, and (2) polyps have the diverse scales, locations and shapes with blurred boundary. To address the issues, Dataset-Level Color Augmentation (DLCA) and Convolutional Multi-scale Attention Module (CMAM) are proposed. DLCA employs the dataset-level color knowledge and generate new color appearances, to avoid the model learning false associations with colors. Meanwhile, CMAM simultaneously explores the region and boundary clues and model multi-scale context, which improves the accuracy of polyp localization and fine-grained segmentation. We conduct comprehensive experiments and compare our network with state-of-the-art methods. The proposed model is superior on multiple polyp datasets, especially on ETIS, where mDice and mIoU reach 0.839 and 0.766. Furthermore, it is validated that DLCA can be widely applied to most polyp segmentation methods, and CMAM is practical and plug-and-play. Finally, the model demonstrates promising generalization to breast ultrasound and skin lesion segmentation.

Non-destructive testing based on Unet-CBAM network for pulsed thermography

Infrared thermography (IRT) is a non-destructive testing technique that can detect the internal defects of materials. In the detection of austenitic stainless-steel pipes with large curvature, image noise caused by uneven heating is difficult to avoid. Traditional image processing methods are less effective. According to previous works, a supervised neural network was proposed in this paper using Unet network and convolutional block attention module. Existing image processing method and networks were used to compare with the proposed method. The results show that the proposed method can remove the noise caused by uneven heating, and detect all subsurface defects in stainless-steel pipe.

YOLOv8-E: An Improved YOLOv8 Algorithm for Eggplant Disease Detection

During the developmental stages, eggplants are susceptible to diseases, which can impact crop yields and farmers’ economic returns. Therefore, timely and effective detection of eggplant diseases is crucial. Deep learning-based object detection algorithms can automatically extract features from images of eggplants affected by diseases. However, eggplant disease images captured in complex farmland environments present challenges such as varying disease sizes, occlusion, overlap, and small target detection, making it difficult for existing deep-learning models to achieve satisfactory detection performance. To address this challenge, this study proposed an optimized eggplant disease detection algorithm, YOLOv8-E, based on You Only Look Once version 8 nano (YOLOv8n). Firstly, we integrate switchable atrous convolution (SAConv) into the C2f module to design the C2f_SAConv module, replacing some of the C2f modules in the backbone network of YOLOv8n, enabling our proposed algorithm to better extract eggplant disease features. Secondly, to facilitate the deployment of the detection model on mobile devices, we reconstruct the Neck network of YOLOv8n using the SlimNeck module, making the model lighter. Additionally, to tackle the issue of missing small targets, we embed the large separable kernel attention (LSKA) module within SlimNeck, enhancing the model’s attention to fine-grained information. Lastly, we combined intersection over union with auxiliary bounding box (Inner-IoU) and minimum point distance intersection over union (MPDIoU), introducing the Inner-MPDIoU loss to speed up convergence of the model and raise detection precision of overlapped and occluded targets. Ablation studies demonstrated that, compared to YOLOv8n, the mean average precision (mAP) and F1 score of YOLOv8-E reached 79.4% and 75.7%, respectively, which obtained a 5.5% increment and a 4.5% increase, while also reducing the model size and computational complexity. Furthermore, YOLOv8-E achieved higher detection performance than other mainstream algorithms. YOLOv8-E exhibits significant potential for practical application in eggplant disease detection.

Music source feature extraction based on improved attention mechanism and phase feature

Music source feature extraction is an important research direction in music information retrieval and music recommendation system. To extract the features of music sources more effectively, the study introduces the jump attention mechanism and combines it with the convolutional attention module. Also, a feature extraction module based on Unet + + and spatial attention module is proposed. In addition, the phase feature information of the mixed music signals is utilized to improve the network performance. Results showed that this model was studied to perform well in music source separation experiments of vocals and accompaniment. For vocal separation on the MIR-1K dataset, the model achieves 11.25 dB, 17.34 dB, and 13.83 dB for each metric, respectively. Meanwhile, for drum separation on the DSD100 dataset, the model achieves a median signal-to-source distortion ratio of 4.36 dB, which is 2.91 dB better than that of the Spectral Hierarchical Network model. For the separation of the bass sound and the human voice, the model's in the separation of bass and human voice, the median distortion ratio of the model is as high as 4.87 dB and 6.09 dB, which is better than that of the Spectral Hierarchical Network model. This indicates the significant performance advantages in feature extraction and separation of music sources, and it has important application values in music production and speech recognition.

CBAM-RIUnet: Breast Tumor Segmentation With Enhanced Breast Ultrasound and Test-Time Augmentation.

This study addresses the challenge of precise breast tumor segmentation in ultrasound images, crucial for effective Computer-Aided Diagnosis (CAD) in breast cancer. We introduce CBAM-RIUnet, a deep learning (DL) model for automated breast tumor segmentation in breast ultrasound (BUS) images. The model, featuring an efficient convolutional block attention module residual inception Unet, outperforms existing models, particularly excelling in Dice and IoU scores. CBAM-RIUnet follows the Unet structure with a residual inception depth-wise separable convolution, and incorporates a convolutional block attention module (CBAM) to eliminate irrelevant features and focus on the region of interest. Evaluation under three scenarios, including enhanced breast ultrasound (EBUS) and test-time augmentation (TTA), demonstrates impressive results. CBAM-RIUnet achieves Dice and IoU scores of 89.38% and 88.71%, respectively, showcasing significant improvements compared to state-of-the-art DL techniques. In conclusion, CBAM-RIUnet presents a highly effective and simplified DL model for breast tumor segmentation in BUS imaging.

MobilenetV2-RC: A lightweight network model for retinopathy classification in retinal OCT images

Abstract Retinopathy is an important ophthalmic disease that causes blindness in the elderly population. With the global old peoples increasing, the ophthalmic healthcare system is more and significant to pre-diagnosis. Optical Coherence Tomography (OCT) is considered the gold standard for ophthalmic treatment and diagnosis. OCT technologies and equipment continue to develop towards the intelligence and convenience for requirements of rapid diagnosis in the remote and poverty-stricken areas. We proposed an improved MobilenetV2 lightweight model for retinopathy classification (MobilenetV2-RC), which incorporates Spatial and Channel Reconstruction Convolution (SCConv) and the improved Convolutional Block Attention Module (CBAM) attention mechanism into the framework. Not only can it effectively limit feature redundancy to reduce model parameters, but also enhance the ability of feature representation to improve classification accuracy. The parameters of the proposed model are only 1.96 M with an overall accuracy of 98.96%, which is higher 3.32% than the original MobilenetV2. Compared with ResNet18, InceptionV3, and VGG16_BN, the overall accuracy is increased by 4.6%, 6.3%, and 3.9%, respectively. The test results of UCSD and Duke open-source datasets are more remarkable. Experimental results show that our proposed algorithm has strong reliability and generalization for the accurate classification of retinopathy, and a greater application prospect in the intelligent diagnosis of ophthalmology and mobile detection terminals.