Search Results (2,986)

Diabetic macular edema (DME) is a leading cause of vision impairment in diabetic patients, necessitating a timely and accurate diagnosis. This paper proposes a comprehensive system for DME grading using retinal fundus images. Our approach integrates multiple deep learning modules, each designed to address key aspects of the diagnostic process. The first module employs the ConvUNeXt model for segmenting hard exudates (HaEx), crucial indicators of DME. The second module uses RetinaNet for precise optic disc (OD) localization, which is essential for subsequent macula localization. The third module refines macula localization, leveraging preprocessing techniques to enhance image clarity. Finally, our system consolidates these results to provide interpretable DME grading. Experimental evaluations were conducted on the Messidor dataset, demonstrating the system’s robust performance. The HaEx segmentation module achieved a mean IoU of 70.5% and a Dice coefficient of 0.64. The OD localization module showed perfect accuracy, recall, and precision at 1.0. For macula localization, our method satisfied the 1R criterion with 99.38% accuracy. The DME grading module achieved an overall accuracy of 91.12%, with an AUC of 0.9334. Our method offers a balanced performance across accuracy, sensitivity, and specificity compared to other non-interpretable and partially interpretable methods. Full article

During the vegetation growing season, the forest in the remote sensing image is more distinguishable from other background features, and the forest features are obvious and can show prominent forest area characteristics. However, deep convolutional neural network-based methods tend to overlearn the forest features in the forest extraction task, which leads to the extraction speed still having a large amount of room for improvement. In this paper, a convolutional neural network-based model is proposed based on the incorporation of spatial and channel reconstruction convolution in the U-Net model for forest extraction from remote sensing images. The network obtained an extraction accuracy of 81.781% in intersection over union (IoU), 91.317% in precision, 92.177% in recall, and 91.745% in F1-score, with a maximum improvement of 0.442% in precision when compared with the classical U-Net network. In addition, the speed of the model’s forest extraction has been improved by about 6.14 times. On this basis, we constructed a forest land dataset with high-intraclass diversity and fine-grained scale by selecting some Sentinel-2 images in Northeast China. The spatial and temporal evolutionary changes of the forest cover in the Fuxin region of Liaoning province, China, from 2019 to 2023, were obtained using this region as the study area. In addition, we obtained the change of the forest landscape pattern evolution in the Fuxin region from 2019 to 2023 based on the morphological spatial pattern analysis (MSPA) method. The results show that the core area of the forest landscape in the Fuxin region has shown an increasing change, and the non-core area has been decreasing. The SC-UNet method proposed in this paper can realize the high-precision and rapid extraction of forest in a wide area, and at the same time, it can provide a basis for evaluating the effectiveness of ecosystem restoration projects. Full article

Accurately detecting tea bud leaves is crucial for the automation of tea picking robots. However, challenges arise due to tea stem occlusion and overlapping of buds and leaves, presenting varied shapes of one bud–one leaf targets in the field of view, making precise segmentation of tea bud leaves challenging. To improve the segmentation accuracy of one bud–one leaf targets with different shapes and fine granularity, this study proposes a novel semantic segmentation model for tea bud leaves. The method designs a hierarchical Transformer block based on a self-attention mechanism in the encoding network, which is beneficial for capturing long-range dependencies between features and enhancing the representation of common features. Then, a multi-path feature aggregation module is designed to effectively merge the feature outputs of encoder blocks with decoder outputs, thereby alleviating the loss of fine-grained features caused by downsampling. Furthermore, a refined polarized attention mechanism is employed after the aggregation module to perform polarized filtering on features in channel and spatial dimensions, enhancing the output of fine-grained features. The experimental results demonstrate that the proposed Unet-Enhanced model achieves segmentation performance well on one bud–one leaf targets with different shapes, with a mean intersection over union (mIoU) of 91.18% and a mean pixel accuracy (mPA) of 95.10%. The semantic segmentation network can accurately segment tea bud leaves, providing a decision-making basis for the spatial positioning of tea picking robots. Full article

The detailed characterization of the pore structure in sandstone is pivotal for the assessment of reservoir properties and the efficiency of oil and gas exploration. Traditional fully supervised learning algorithms are limited in performance enhancement and require a substantial amount of accurately annotated data, which can be challenging to obtain. To address this, we introduce a semi-supervised framework with a U-Net backbone network. Our dataset was curated from 295 two-dimensional CT grayscale images, selected at intervals from nine 4 mm sandstone core samples. To augment the dataset, we employed StyleGAN2-ADA to generate a large number of images with a style akin to real sandstone images. This approach allowed us to generate pseudo-labels through semi-supervised learning, with only a small subset of the data being annotated. The accuracy of these pseudo-labels was validated using ensemble learning methods. The experimental results demonstrated a pixel accuracy of 0.9993, with a pore volume discrepancy of just 0.0035 compared to the actual annotated data. Furthermore, by reconstructing the three-dimensional pore structure of the sandstone, we have shown that the synthetic three-dimensional pores can effectively approximate the throat length distribution of the real sandstone pores and exhibit high precision in simulating throat shapes. Full article

As a core component of electric vehicles, the state of health (SOH) of lithium-ion battery has a direct impact on vehicle performance and safety. Existing data-driven models primarily focus on feature extraction, often overlooking the processing of multi-level redundant information and the utilization of multi-stage battery features. To address the issues, this paper proposes a novel data-driven method, named multi-stage mixed attention U-Net (MMAU-Net), for SOH estimation, which is based on both the phased learning and an encoder–decoder structure. First, the geometric knee-point division method is proposed to divide the battery life cycle into multiple stages, which allows the model to learn the distinctive features of battery degradation at each stage. Second, to adeptly capture degraded features and reduce redundant information, we propose a mixed attention U-Net model for the SOH prediction task, which is constructed upon the fundamental U-Net backbone and is enhanced with time step attention and feature attention modules. The experimental results validate the proposed method’s feasibility and efficacy, demonstrating an acceptable performance across a spectrum of evaluative metrics. Consequently, this study offers a research within the domain of battery health management. Full article

In this work, we enhanced the U-shaped network and proposed a method for detecting scratches on metal surfaces based on the Metal Surface Defect Detection U-Net (MSDD-UNet). Initially, we integrated a downsampling approach using a Space-To-Depth module and a lightweight channel attention module to address the loss of contextual information in feature maps that results from multiple convolution and pooling operations. Building on this, we developed an improved attention module that utilizes image frequency decomposition and cross-channel self-attention mechanisms, as well as the strengths of convolutional encoders and self-attention blocks. Additionally, this attention module was integrated into the skip connections between the encoder and decoder. The purpose was to capture dense contextual information, highlight small and fine target areas, and assist in localizing micro and fine scratch defects. In response to the severe foreground–background class imbalance in scratch images, a hybrid loss function combining focal loss and $D_{ice}$ loss was put forward to train the model for precise scratch segmentation. Finally, experiments were conducted on two surface defect datasets. The results reveal that our proposed method is more advantageous than other state-of-the-art scratch segmentation methods. Full article

Camellia oleifera is a crop of high economic value, yet it is particularly susceptible to various diseases and pests that significantly reduce its yield and quality. Consequently, the precise segmentation and classification of diseased Camellia leaves are vital for managing pests and diseases effectively. Deep learning exhibits significant advantages in the segmentation of plant diseases and pests, particularly in complex image processing and automated feature extraction. However, when employing single-modal models to segment Camellia oleifera diseases, three critical challenges arise: (A) lesions may closely resemble the colors of the complex background; (B) small sections of diseased leaves overlap; (C) the presence of multiple diseases on a single leaf. These factors considerably hinder segmentation accuracy. A novel multimodal model, CNN–Transformer Dual U-shaped Network (CTDUNet), based on a CNN–Transformer architecture, has been proposed to integrate image and text information. This model first utilizes text data to address the shortcomings of single-modal image features, enhancing its ability to distinguish lesions from environmental characteristics, even under conditions where they closely resemble one another. Additionally, we introduce Coordinate Space Attention (CSA), which focuses on the positional relationships between targets, thereby improving the segmentation of overlapping leaf edges. Furthermore, cross-attention (CA) is employed to align image and text features effectively, preserving local information and enhancing the perception and differentiation of various diseases. The CTDUNet model was evaluated on a self-made multimodal dataset compared against several models, including DeeplabV3+, UNet, PSPNet, Segformer, HrNet, and Language meets Vision Transformer (LViT). The experimental results demonstrate that CTDUNet achieved an mean Intersection over Union (mIoU) of 86.14%, surpassing both multimodal models and the best single-modal model by 3.91% and 5.84%, respectively. Additionally, CTDUNet exhibits high balance in the multi-class segmentation of Camellia oleifera diseases and pests. These results indicate the successful application of fused image and text multimodal information in the segmentation of Camellia disease, achieving outstanding performance. Full article

Post-mortem (PM) imaging has potential for identifying individuals by comparing ante-mortem (AM) and PM images. Radiographic images of bones contain significant information for personal identification. However, PM images are affected by soft tissue decomposition; therefore, it is desirable to extract only images of bones that change little over time. This study evaluated the effectiveness of U-Net for bone image extraction from two-dimensional (2D) X-ray images. Two types of pseudo 2D X-ray images were created from the PM computed tomography (CT) volumetric data using ray-summation processing for training U-Net. One was a projection of all body tissues, and the other was a projection of only bones. The performance of the U-Net for bone extraction was evaluated using Intersection over Union, Dice coefficient, and the area under the receiver operating characteristic curve. Additionally, AM chest radiographs were used to evaluate its performance with real 2D images. Our results indicated that bones could be extracted visually and accurately from both AM and PM images using U-Net. The extracted bone images could provide useful information for personal identification in forensic pathology. Full article

Landslide disasters pose significant threats to human life and property; therefore, accurate and effective detection and area extraction methods are crucial in environmental monitoring and disaster management. In our study, we address the critical tasks of landslide detection and area extraction in remote sensing images using advanced deep learning techniques. For landslide detection, we propose an enhanced dual-channel model that leverages EfficientNetB7 for feature extraction and incorporates spatial attention mechanisms (SAMs) to enhance important features. Additionally, we utilize a deep separable convolutional neural network with a Transformers module for feature extraction from digital elevation data (DEM). The extracted features are then fused using a variational autoencoder (VAE) to mine potential features and produce final classification results. Experimental results demonstrate impressive accuracy rates of 98.92% on the Bijie City landslide dataset and 94.70% on the Landslide4Sense dataset. For landslide area extraction, we enhance the traditional Unet++ architecture by incorporating Dilated Convolution to expand the receptive field and enable multi-scale feature extraction. We further integrate the Transformer and Convolutional Block Attention Module to enhance feature focus and introduce multi-task learning, including segmentation and edge detection tasks, to efficiently extract and refine landslide areas. Additionally, conditional random fields (CRFs) are applied for post-processing to refine segmentation boundaries. Comparative analysis demonstrates the superior performance of our proposed model over traditional segmentation models such as Unet, Fully Convolutional Network (FCN), and Segnet, as evidenced by improved metrics: IoU of 0.8631, Dice coefficient of 0.9265, overall accuracy (OA) of 91.53%, and Cohen’s kappa coefficient of 0.9185 on the Bijie City landslide dataset; and IoU of 0.8217, Dice coefficient of 0.9021, overall accuracy (OA) of 96.68%, and Cohen’s kappa coefficient of 0.8835 on the Landslide4Sense dataset. These findings highlight the effectiveness and robustness of our proposed methodologies in addressing critical challenges in landslide detection and area extraction tasks, with significant implications for enhancing disaster management and risk assessment efforts in remote sensing applications. Full article

Thermal infrared imaging could detect hidden faults in various types of high-voltage power equipment, which is of great significance for power inspections. However, there are still certain issues with thermal-imaging-based abnormal heating detection methods due to varying appearances of abnormal regions and complex temperature interference from backgrounds. To solve these problems, a contour-based instance segmentation network is first proposed to utilize thermal (T) and visual (RGB) images, realizing high-accuracy segmentation against complex and changing environments. Specifically, modality-specific features are encoded via two-stream backbones and fused in spatial, channel, and frequency domains. In this way, modality differences are well handled, and effective complementary information is extracted for object detection and contour initialization. The transformer decoder is further utilized to explore the long-range relationships between contour points with background points, and to achieve the deformation of contour points. Then, the auto-encoder-based reconstruction network is developed to learn the distribution of power equipment using the proposed random argument strategy. Meanwhile, the UNet-like discriminative network directly explores the differences between the reconstructed and original image, capturing the deviation of poor reconstruction regions for abnormal heating detection. Many images are acquired in transformer substations with different weathers and day times to build the datasets with pixel-level annotation. Several extensive experiments are conducted for qualitative and quantitative evaluation, while the comparison results fully prove the effectiveness and robustness of the proposed instance segmentation method. The practicality and performance of the proposed abnormal heating detection method are evaluated on image patches with different kinds of insulators. Full article

Distracted driving is one of the major factors leading drivers to ignore potential road hazards. In response to the challenges of high computational complexity, limited generalization capacity, and suboptimal detection accuracy in existing deep learning-based detection algorithms, this paper introduces a novel approach called StarDL-YOLO (StarNet-detectlscd-yolo), which leverages an enhanced version of YOLOv8n. Initially, the StarNet integrated into the backbone of YOLOv8n significantly improves the feature extraction capability of the model with remarkable reduction in computational complexity. Subsequently, the Star Block is incorporated into the neck network, forming a C2f-Star module that offers lower computational cost. Additionally, shared convolution is introduced in the detection head to further reduce computational burden and parameter size. Finally, the Wise-Focaler-MPDIoU loss function is proposed to strengthen detection accuracy. The experimental results demonstrate that StarDL-YOLO significantly improves the efficiency of the distracted driving behavior detection, achieving an accuracy of 99.6% on the StateFarm dataset. Moreover, the parameter count of the model is minimized by 56.4%, and its computational load is decreased by 45.1%. Additionally, generalization experiments are performed on the 100-Driver dataset, revealing that the proposed scheme enhances generalization effectiveness compared to YOLOv8n. Therefore, this algorithm significantly reduces computational load while maintaining high reliability and generalization capability. Full article

Currently, existing deep learning methods exhibit many limitations in multi-target detection, such as low accuracy and high rates of false detection and missed detections. This paper proposes an improved Faster R-CNN algorithm, aiming to enhance the algorithm’s capability in detecting multi-scale targets. This algorithm has three improvements based on Faster R-CNN. Firstly, the new algorithm uses the ResNet101 network for feature extraction of the detection image, which achieves stronger feature extraction capabilities. Secondly, the new algorithm integrates Online Hard Example Mining (OHEM), Soft non-maximum suppression (Soft-NMS), and Distance Intersection Over Union (DIOU) modules, which improves the positive and negative sample imbalance and the problem of small targets being easily missed during model training. Finally, the Region Proposal Network (RPN) is simplified to achieve a faster detection speed and a lower miss rate. The multi-scale training (MST) strategy is also used to train the improved Faster R-CNN to achieve a balance between detection accuracy and efficiency. Compared to the other detection models, the improved Faster R-CNN demonstrates significant advantages in terms of [email protected], F1-score, and Log average miss rate (LAMR). The model proposed in this paper provides valuable insights and inspiration for many fields, such as smart agriculture, medical diagnosis, and face recognition. Full article

The parametrisation of the geometry in shape optimisation has an important influence on the quality of the optimum and the rate of convergence of the optimiser. Refinement studies for the parametrisation are not shown in the literature, as most methods use non-orthogonal parametrisations, which cause issues with convergence when the parametrisation is refined. The NURBS-based parametrisation with complex constraints (NSPCC) is the only CAD-based parametrisation method that guarantees orthogonal shape modes by constructing an optimal basis. We conduct a parametrisation refinement study for the benchmark turbomachinery cooling bend (U-bend) geometry, an intially symmetric geometry. Using an adjoint RANS solver, we optimise for mininmal total pressure drop. The results show significant effects of the control net density on the final shape, with the finest control net producing an asymmetric optimal shape resembling strakes that induces swirl ahead of the bend. These asymmetric modes have not been reported in the literature so far. We also demonstrate that the convergence rate of the optimiser is not significantly affected by the refinement of the parametrisation. The effectiveness of these shape features obtained with single-point optimisation is evaluated for a range of Reynolds numbers. It is shown that total pressure drop reduction is not sensitive to Reynolds number. Full article

Assessing glaciers using recent and historical data and predicting the future impacts on them due to climate change are crucial for understanding global glacier mass balance, regional water resources, and downstream hydrology. Computational methods are crucial for analyzing current conditions and forecasting glacier changes using remote sensing and other data sources. Due to the complexity and large data volumes, there is a strong demand for accelerated computing. AI-based approaches are increasingly being adopted for their efficiency and accuracy in these tasks. Thus, in the current state-of-the-art review work, available research results on the application of AI methods for glacier studies are addressed. Using selected search terms, AI-based publications are collected from research databases. They are further classified in terms of their geographical locations and glacier-related research purposes. It was found that the majority of AI-based glacier studies focused on inventorying and mapping glaciers worldwide. AI techniques like U-Net, Random forest, CNN, and DeepLab are mostly utilized in glacier mapping, demonstrating their adaptability and scalability. Other AI-based glacier studies such as glacier evolution, snow/ice differentiation, and ice dynamic modeling are reviewed and classified, Overall, AI methods are predominantly based on supervised learning and deep learning approaches, and these methods have been used almost evenly in glacier publications over the years since the beginning of this research area. Thus, the integration of AI in glacier research is advancing, promising to enhance our comprehension of glaciers amid climate change and aiding environmental conservation and resource management. Full article

Synthetic aperture radar (SAR) is a technique widely used in the field of ship detection. However, due to the high ship density, fore-ground-background imbalance, and varying target sizes, achieving lightweight and high-precision multiscale ship object detection remains a significant challenge. In response to these challenges, this research presents YOLO-MSD, a multiscale SAR ship detection method. Firstly, we propose a Deep Poly Kernel Backbone Network (DPK-Net) that utilizes the Optimized Convolution (OC) Module to reduce data redundancy and the Poly Kernel (PK) Module to improve the feature extraction capability and scale adaptability. Secondly, we design a BiLevel Spatial Attention Module (BSAM), which consists of the BiLevel Routing Attention (BRA) and the Spatial Attention Module. The BRA is first utilized to capture global information. Then, the Spatial Attention Module is used to improve the network’s ability to localize the target and capture high-quality detailed information. Finally, we adopt a Powerful-IoU (P-IoU) loss function, which can adjust to the ship size adaptively, effectively guiding the anchor box to achieve faster and more accurate detection. Using HRSID and SSDD as experimental datasets, mAP of 90.2% and 98.8% are achieved, respectively, outperforming the baseline by 5.9% and 6.2% with a model size of 12.3 M. Furthermore, the network exhibits excellent performance across various ship scales. Full article