Search Results (570)

Compared to other multi-source image fusion tasks, visible and SAR image fusion faces a lack of training data in deep learning-based methods. Introducing structural priors to design fusion networks is a viable solution. We incorporated the feature hierarchy concept from computer vision, dividing deep features into low-, mid-, and high-level tiers. Based on the complementary modal characteristics of SAR and visible, we designed a fusion architecture that fully analyze and utilize the difference of hierarchical features. Specifically, our framework has two stages. In the cross-modal enhancement stage, a CycleGAN generator-based method for cross-modal interaction and input data enhancement is employed to generate pseudo-modal images. In the fusion stage, we have three innovations: (1) We designed feature extraction branches and fusion strategies differently for each level based on the features of different levels and the complementary modal features of SAR and visible to fully utilize cross-modal complementary features. (2) We proposed the Layered Strictly Nested Framework (LSNF), which emphasizes hierarchical differences and uses hierarchical characteristics, to reduce feature redundancy. (3) Based on visual saliency theory, we proposed a Gradient-weighted Pixel Loss (GWPL), which dynamically assigns higher weights to regions with significant gradient magnitudes, emphasizing high-frequency detail preservation during fusion. Experiments on the YYX-OPT-SAR and WHU-OPT-SAR datasets show that our method outperforms 11 state-of-the-art methods. Ablation studies confirm each component’s contribution. This framework effectively meets remote sensing applications’ high-precision image fusion needs. Full article

Generalising deep-learning models to perform well on unseen data domains with minimal retraining remains a significant challenge in computer vision. Even when the target task—such as quantifying the number of elements in an image—stays the same, data quality, shape, or form variations can deviate from the training conditions, often necessitating manual intervention. As a real-world industry problem, we aim to automate stock level estimation in retail cabinets. As technology advances, new cabinet models with varying shapes emerge alongside new camera types. This evolving scenario poses a substantial obstacle to deploying long-term, scalable solutions. To surmount the challenge of generalising to new cabinet models and cameras with minimal amounts of sample images, this research introduces a new solution. This paper proposes a novel ensemble model that combines DenseNet-201 and Vision Transformer (ViT-B/8) architectures to achieve generalisation in stock-level classification. The novelty aspect of our solution comes from the fact that we combine a transformer with a DenseNet model in order to capture both the local, hierarchical details and the long-range dependencies within the images, improving generalisation accuracy with less data. Key contributions include (i) a novel DenseNet-201 + ViT-B/8 feature-level fusion, (ii) an adaptation workflow that needs only two images per class, (iii) a balanced layer-unfreezing schedule, (iv) a publicly described domain-shift benchmark, and (v) a 47 pp accuracy gain over four standard few-shot baselines. Our approach leverages fine-tuning techniques to adapt two pre-trained models to the new retail cabinets (i.e., standing or horizontal) and camera types using only two images per class. Experimental results demonstrate that our method achieves high accuracy rates of 91% on new cabinets with the same camera and 89% on new cabinets with different cameras, significantly outperforming standard few-shot learning methods. Full article

The detection of aerial imagery captured by Unmanned Aerial Vehicles (UAVs) is widely employed across various domains, including engineering construction, traffic regulation, and precision agriculture. However, aerial images are typically characterized by numerous small targets, significant occlusion issues, and densely clustered targets, rendering traditional detection algorithms largely ineffective for such imagery. This work proposes a small target detection algorithm, SR-YOLO. It is specifically tailored to address these challenges in UAV-captured aerial images. First, the Space-to-Depth layer and Receptive Field Attention Convolution are combined, and the SR-Conv module is designed to replace the Conv module within the original backbone network. This hybrid module extracts more fine-grained information about small target features by converting image spatial information into depth information and the attention of the network to targets of different scales. Second, a small target detection layer and a bidirectional feature pyramid network mechanism are introduced to enhance the neck network, thereby strengthening the feature extraction and fusion capabilities for small targets. Finally, the model’s detection performance for small targets is improved by utilizing the Normalized Wasserstein Distance loss function to optimize the Complete Intersection over Union loss function. Empirical results demonstrate that the SR-YOLO algorithm significantly enhances the precision of small target detection in UAV aerial images. Ablation experiments and comparative experiments are conducted on the VisDrone2019 and RSOD datasets. Compared to the baseline algorithm YOLOv8s, our SR-YOLO algorithm has improved mAP@0.5 by 6.3% and 3.5% and mAP@0.5:0.95 by 3.8% and 2.3% on the datasets VisDrone2019 and RSOD, respectively. It also achieves superior detection results compared to other mainstream target detection methods. Full article

Breast cancer remains one of the leading causes of mortality among women worldwide, highlighting the critical need for early and accurate detection methods. Traditional mammography, although widely used, has limitations, including radiation exposure and challenges in detecting early-stage lesions. Electrical Impedance Mammography (EIM) has emerged as a non-invasive and radiation-free alternative that assesses the density and electrical conductivity of breast tissue. EIM images consist of seven layers, each representing different tissue depths, offering a detailed representation of the breast structure. However, analyzing these layers individually can be redundant and complex, making it difficult to identify relevant features for lesion classification. To address this issue, advanced computational techniques are employed for image integration, such as the Root Mean Square ($C_{\mathrm{RMS}}$) Contrast and Contrast-Limited Adaptive Histogram Equalization (CLAHE), combined with the Coefficient of Variation (CV), CLAHE-based fusion, weighted average fusion, Gaussian pyramid fusion, and Wavelet–PCA fusion. Each method enhances the representation of tissue features, optimizing the image quality and diagnostic utility. This study evaluated the impact of these integration techniques on EIM image analysis, aiming to improve the accuracy and reliability of computational diagnostic models for breast cancer detection. According to the obtained results, the best performance was achieved using Wavelet–PCA fusion in combination with XGBoost as a classifier, yielding an accuracy rate of 89.5% and an F1-score of 81.5%. These results are highly encouraging for the further investigation of this topic. Full article

The selection of appropriate biomarkers in clinical practice aids in the early detection, treatment, and prevention of disease while also assisting in the development of targeted therapeutics. Recently, multi-omics data generated from advanced technology platforms has become available for disease studies. Therefore, the integration of this data with associated clinical data provides a unique opportunity to gain a deeper understanding of disease. However, the effective integration of large-scale multi-omics data remains a major challenge. To address this, we propose a novel deep learning model—the Multi-Omics Graph Attention biomarker Discovery network (MOGAD). MOGAD aims to efficiently classify diseases and discover biomarkers by integrating various omics data such as DNA methylation, gene expression, and miRNA expression. The model consists of three main modules: Multi-head GAT network (MGAT), Multi-Graph Attention Fusion (MGAF), and Attention Fusion (AF), which work together to dynamically model the complex relationships among different omics layers. We incorporate clinical data (e.g., APOE genotype) which enables a systematic investigation of the influence of non-omics factors on disease classification. The experimental results demonstrate that MOGAD achieves a superior performance compared to existing single-omics and multi-omics integration methods in classification tasks for Alzheimer’s disease (AD). In the comparative experiment on the ROSMAP dataset, our model achieved the highest ACC (0.773), F1-score (0.787), and MCC (0.551). The biomarkers identified by MOGAD show strong associations with the underlying pathogenesis of AD. We also apply a Hi-C dataset to validate the biological rationality of the identified biomarkers. Furthermore, the incorporation of clinical data enhances the model’s robustness and uncovers synergistic interactions between omics and non-omics features. Thus, our deep learning model is able to successfully integrate multi-omics data to efficiently classify disease and discover novel biomarkers. Full article

Change detection (CD) is essential for Earth observation tasks, as it identifies alterations in specific geographic areas over time. The advancement of deep learning has significantly improved the accuracy of CD. However, encoder–decoder architectures often struggle to effectively capture temporal differences. Encoders may lose critical spatial details, while decoders can introduce bias due to inconsistent receptive fields across layers. To address these limitations, this paper proposes an enhanced symmetric network, termed FDRF (feature difference and receptive field), which incorporates two novel components: the multibranch feature difference extraction (MFDE) module and the adaptive ensemble decision (AED) module. MFDE independently extracts differential features from bitemporal images at each encoder layer, using multiscale fusion to retain image content and improve the quality of feature difference modeling. AED assigns confidence weights to predictions from different decoder layers based on their receptive field sizes and then combines them adaptively to reduce scale-related bias. To validate the effectiveness and robustness of FDRF, experiments were conducted on five public datasets (SYSU, LEVIR-CD, WHU, NJDS, and CLCD), as well as a UAV-based dataset collected from two national coastal nature reserves in Guangxi Beihai, China. The results demonstrate that FDRF consistently outperforms existing methods in accuracy and robustness across diverse scenarios. Full article

Within the field of remote sensing, Unmanned Aerial Vehicle (UAV) infrared object detection plays a pivotal role, especially in complex environments. However, existing methods face challenges such as insufficient accuracy or low computational efficiency, particularly in the detection of small objects. This paper proposes a lightweight and accurate UAV infrared object detection model, YOLO-UIR, for small object detection from a UAV perspective. The model is based on the YOLO architecture and mainly includes the Efficient C2f module, lightweight spatial perception (LSP) module, and bidirectional feature interaction fusion (BFIF) module. The Efficient C2f module significantly enhances feature extraction capabilities by combining local and global features through an Adaptive Dual-Stream Attention Mechanism. Compared with the existing C2f module, the introduction of Partial Convolution reduces the model’s parameter count while maintaining high detection accuracy. The BFIF module further enhances feature fusion effects through cross-level semantic interaction, thereby improving the model’s ability to fuse contextual features. Moreover, the LSP module efficiently combines features from different distances using Large Receptive Field Convolution Layers, significantly enhancing the model’s long-range information capture capability. Additionally, the use of Reparameterized Convolution and Depthwise Separable Convolution ensures the model’s lightweight nature, making it highly suitable for real-time applications. On the DroneVehicle and HIT-UAV datasets, YOLO-UIR achieves superior detection performance compared to existing methods, with an mAP of 71.1% and 90.7%, respectively. The model also demonstrates significant advantages in terms of computational efficiency and parameter count. Ablation experiments verify the effectiveness of each optimization module. Full article

Printed circuit board (PCB) defect detection faces challenges like small target feature loss and severe background interference. To address these issues, this paper proposes PCES-YOLO, an enhanced YOLOv11-based model. First, a developed Pre-convolution Receptive Field Enhancement (PRFE) module replaces C3k in the C3k2 module. The ConvNeXtBlock with inverted bottleneck is introduced in the P4 layer, greatly improving small-target feature capture and semantic understanding. The second key innovation lies in the creation of the Efficient Feature Fusion and Aggregation Network (EFAN), which integrates a lightweight Spatial-Channel Decoupled Downsampling (SCDown) module and three innovative fusion pathways. This achieves substantial parameter reduction while effectively integrating shallow detail features with deep semantic features, preserving critical defect information across different feature levels. Finally, the Shape-IoU loss function is incorporated, focusing on bounding box shape and scale for more accurate regression and enhanced defect localization precision. Experiments on the enhanced Peking University PCB defect dataset show that PCES-YOLO achieves a mAP50 of 97.3% and a mAP50–95 of 77.2%. Compared to YOLOv11n, it shows improvements of 3.6% in mAP50 and 15.2% in mAP50–95. When compared to YOLOv11s, it increases mAP50 by 1.0% and mAP50–95 by 5.6% while also significantly reducing the model parameters. The performance of PCES-YOLO is also evaluated against mainstream object detection algorithms, including Faster R-CNN, SSD, YOLOv8n, etc. These results indicate that PCES-YOLO outperforms these algorithms in terms of detection accuracy and efficiency, making it a promising high-precision and efficient solution for PCB defect detection in industrial settings. Full article

Underwater sonar target recognition is crucial in fields such as national defense, navigation, and environmental monitoring. However, it faces issues such as the complex characteristics of ship-radiated noise, imbalanced data distribution, non-stationarity, and bottlenecks of existing technologies. This paper proposes the MultiFuseNet-AID network, aiming to address these challenges. The network includes the TriFusion block module, the novel lightweight attention residual network (NLARN), the long- and short-term attention (LSTA) module, and the Mamba module. Through the TriFusion block module, the original, differential, and cumulative signals are processed in parallel, and features such as MFCC, CQT, and Fbank are fused to achieve deep multi-domain feature fusion, thereby enhancing the signal representation ability. The NLARN was optimized based on the ResNet architecture, with the SE attention mechanism embedded. Combined with the long- and short-term attention (LSTA) and the Mamba module, it could capture long-sequence dependencies with an O(N) complexity, completing the optimization of lightweight long sequence modeling. At the same time, with the help of feature fusion, and layer normalization and residual connections of the Mamba module, the adaptability of the model in complex scenarios with imbalanced data and strong noise was enhanced. On the DeepShip and ShipsEar datasets, the recognition rates of this model reached 98.39% and 99.77%, respectively. The number of parameters and the number of floating point operations were significantly lower than those of classical models, and it showed good stability and generalization ability under different sample label ratios. The research shows that the MultiFuseNet-AID network effectively broke through the bottlenecks of existing technologies. However, there is still room for improvement in terms of adaptability to extreme underwater environments, training efficiency, and adaptability to ultra-small devices. It provides a new direction for the development of underwater sonar target recognition technology. Full article

Aiming at the problem of appearance drift and susceptibility to noise interference in visual place recognition (VPR), we propose DD–DPFE: a Dynamic Difference and Dual-Path Feature Enhancement method. Embedding differential attention mechanisms in the DINOv2 model to mitigate the effects of process interference and adding serial-parallel adapters allows efficient model parameter migration and task adaptation. Our method constructs a two-way feature enhancement module with global–local branching synergy. The global branch employs a dynamic fusion mechanism with a multi-layer Transformer encoder to strengthen the structured spatial representation to cope with appearance changes, while the local branch suppresses the over-response of redundant noise through an adaptive weighting mechanism and fuses the contextual information from the multi-scale feature aggregation module to enhance the robustness of the scene. The experimental results show that the model architecture proposed in this paper is an obvious improvement in different environmental tests. This is most obvious in the simulation test of a night scene, verifying that the proposed method can effectively enhance the discriminative power of the system and its anti-jamming ability in complex scenes. Full article

Entity alignment is a critical technique for integrating diverse knowledge graphs. Although existing methods have achieved impressive success in traditional entity alignment, they may struggle to handle the complexities arising from interactions and dependencies in multi-modal knowledge. In this paper, a novel multi-modal entity alignment model called ERMF is proposed, which leverages distinct modal characteristics of entities to identify equivalent entities across different multi-modal knowledge graphs. The symmetry in cross-modal interactions and hierarchical feature fusion is a core design principle of our approach. Specifically, we first utilize different feature encoders to independently extract features from different modalities. Concurrently, visual features and nearest neighbor negative sampling methods are incorporated to design a vision-guided negative sample generation strategy based on contrastive learning, ensuring a symmetric balance between positive and negative samples and guiding the model to learn effective relationship embeddings. Subsequently, in the feature fusion stage, we propose a multi-layer feature fusion approach that incorporates cross-attention and cross-modal attention mechanisms with symmetric processing of intra- and inter-modal correlations, thereby obtaining multi-granularity features. Extensive experiments were conducted on two public datasets, namely FB15K-DB15K and FB15K-YAGO15K. With 20% aligned seeds, ERMF improves Hits@1 by 8.4% and 26%, and MRR by 6% and 19.2% compared to the best baseline. The symmetric architecture of our model ensures the robust and balanced utilization of multi-modal information, aligning with the principles of structural and functional symmetry in knowledge integration. Full article

The poor surface quality of the metal parts produced by laser powder bed fusion limits their application in load-bearing components, as it promotes crack initiation under cyclic loadings. Consequently, improving part quality relies on time-consuming surface finishing. This work explores a dual-laser powder bed fusion strategy to simultaneously improve the productivity, surface quality, and fatigue life of parts with inclined up-facing surfaces made from a novel tool steel. This is achieved by combining building using a high layer thickness of 120 μm with in situ quality enhancement through powder removal and laser remelting. A bending fatigue campaign was conducted to assess the performance of such treated samples produced with different layer thicknesses (60 μm, hull-bulk 60/120 μm, 120 μm) compared to as-built and machined reference samples. Remelting consistently enhanced the fatigue life compared to the as-built reference samples by up to a factor of 36. The improvement was attributed to the reduced surface roughness, the reduced critical stress concentration factors, and the gradually changing surface features with increased lateral dimensions. This led to a beneficial load distribution and fewer potential crack initiation points. Finally, the remelting samples produced with a layer thickness of 120 μm enhanced the fatigue life by a factor of four and reduced the production time by 30% compared to the standard approach using a layer thickness of 60 μm. Full article

Recently, STDenseNet (SpatioTemporal Densely connected convolutional Network) showed remarkable performance in predicting network traffic by leveraging the inductive bias of convolution layers. However, it is known that such convolution layers can only barely capture long-term spatial and temporal dependencies. To solve this problem, we propose Attention-DenseNet (ADNet), which effectively incorporates an attention module into STDenseNet to learn representations for long-term spatio-temporal patterns. Specifically, we explored the optimal positions and the types of attention modules in combination with STDenseNet. Our key findings are as follows: i) attention modules are very effective when positioned between the last dense module and the final feature fusion module, meaning that the attention module plays a key role in aggregating low-level local features with long-term dependency. Hence, the final feature fusion module can easily exploit both global and local information; ii) the best attention module is different depending on the spatio-temporal characteristics of the dataset. To verify the effectiveness of the proposed ADNet, we performed experiments on the Telecom Italia dataset, a well-known benchmark dataset for network traffic prediction. The experimental results show that, compared to STDenseNet, our ADNet improved RMSE performance by 3.72%, 2.84%, and 5.87% in call service (Call), short message service (SMS), and Internet access (Internet) sub-datasets, respectively. Full article

Single image dehazing is a fundamental task in computer vision, aiming to recover a clear scene from a hazy input image. To address the limitations of traditional dehazing algorithms—particularly in global feature association and local detail preservation—this study proposes a novel Transformer-based dehazing model enhanced by an interactive channel attention mechanism. The proposed architecture adopts a U-shaped encoder–decoder framework, incorporating key components such as a feature extraction module and a feature fusion module based on interactive attention. Specifically, the interactive channel attention mechanism facilitates cross-layer feature interaction, enabling the dynamic fusion of global contextual information and local texture details. The network architecture leverages a multi-scale feature pyramid to extract image information across different dimensions, while an improved cross-channel attention weighting mechanism enhances feature representation in regions with varying haze densities. Extensive experiments conducted on both synthetic and real-world datasets—including the RESIDE benchmark—demonstrate the superior performance of the proposed method. Quantitatively, it achieves PSNR gains of 0.53 dB for indoor scenes and 1.64 dB for outdoor scenes, alongside SSIM improvements of 1.4% and 1.7%, respectively, compared with the second-best performing method. Qualitative assessments further confirm that the proposed model excels in restoring fine structural details in dense haze regions while maintaining high color fidelity. These results validate the effectiveness of the proposed approach in enhancing both perceptual quality and quantitative accuracy in image dehazing tasks. Full article

In recent years, with the rapid development of the Internet and mobile devices, the high-resolution video industry has ushered in a booming golden era, making video content the primary driver of Internet traffic. This trend has spurred continuous innovation in efficient video coding technologies, such as Advanced Video Coding/H.264 (AVC), High Efficiency Video Coding/H.265 (HEVC), and Versatile Video Coding/H.266 (VVC), which significantly improves compression efficiency while maintaining high video quality. However, during the encoding process, compression artifacts and the loss of visual details remain unavoidable challenges, particularly in high-resolution video processing, where the massive amount of image data tends to introduce more artifacts and noise, ultimately affecting the user’s viewing experience. Therefore, effectively reducing artifacts, removing noise, and minimizing detail loss have become critical issues in enhancing video quality. To address these challenges, this paper proposes a post-processing method based on Convolutional Neural Network (CNN) that improves the quality of VVC-reconstructed frames through deep feature extraction and fusion. The proposed method is built upon a high-resolution dual-path residual gating system, which integrates deep features from different convolutional layers and introduces convolutional blocks equipped with gating mechanisms. By ingeniously combining gating operations with residual connections, the proposed approach ensures smooth gradient flow while enhancing feature selection capabilities. It selectively preserves critical information while effectively removing artifacts. Furthermore, the introduction of residual connections reinforces the retention of original details, achieving high-quality image restoration. Under the same bitrate conditions, the proposed method significantly improves the Peak Signal-to-Noise Ratio (PSNR) value, thereby optimizing video coding quality and providing users with a clearer and more detailed visual experience. Extensive experimental results demonstrate that the proposed method achieves outstanding performance across Random Access (RA), Low Delay B-frame (LDB), and All Intra (AI) configurations, achieving BD-Rate improvements of 6.1%, 7.36%, and 7.1% for the luma component, respectively, due to the remarkable PSNR enhancement. Full article