Dense Features Research Articles

Automated Road Extraction from Satellite Imagery Integrating Dense Depthwise Dilated Separable Spatial Pyramid Pooling with DeepLabV3+

Road extraction is a sub-domain of remote sensing applications; it is a subject of extensive and ongoing research. The procedure of automatically extracting roads from satellite imagery encounters significant challenges due to the multi-scale and diverse structures of roads; improvement in this field is needed. Convolutional neural networks (CNNs), especially the DeepLab series known for its proficiency in semantic segmentation due to its efficiency in interpreting multi-scale objects’ features, address some of these challenges caused by the varying nature of roads. The present work proposes the utilization of DeepLabV3+, the latest version of the DeepLab series, by introducing an innovative Dense Depthwise Dilated Separable Spatial Pyramid Pooling (DenseDDSSPP) module and integrating it in the place of the conventional Atrous Spatial Pyramid Pooling (ASPP) module. This modification enhances the extraction of complex road structures from satellite images. This study hypothesizes that the integration of DenseDDSSPP with a CNN backbone network and a Squeeze-and-Excitation block will generate an efficient dense feature map by focusing on relevant features, leading to more precise and accurate road extraction from remote sensing images. The Results Section presents a comparison of our model’s performance against state-of-the-art models, demonstrating better results that highlight the effectiveness and success of the proposed approach.

Dense Feature Interaction Network for Image Inpainting Localization

Dense Feature Interaction Network for Image Inpainting Localization

Improvement of the urban microclimate with aromatic and medicinal plants

The paper examines the bioclimatic contribution of urban gardening through the use of aromatic and medicinal plants, proposing an innovative approach to urban design. The research aims to evaluate the role of these plants in improving the urban microclimate, focusing on their ability to reduce temperature, increase relative humidity, and contribute to the creation of a synergistic ecosystem that attracts beneficial insects. The methodology involved experimental research at eight sites in Thessaloniki, where temperature, humidity, and solar radiation were recorded. At the sites with aromatic and medicinal plants, a temperature reduction of 10-15% was observed compared to the non-planted areas. Significant improvement was recorded in areas simulating balconies, where potted plants led to a temperature reduction of 5-10%. Furthermore, areas with dense vegetation and water features showed greater humidity, while air circulation was limited, creating adverse thermal comfort conditions. In locations with small arrangements of aromatic plants, a reduction in discomfort perception of up to 26% was observed. The study highlights the innovative potential of integrating aromatic and medicinal plants into urban design for improving the microclimate. Replacing grass in public spaces and using these plants on vertical and horizontal surfaces can improve thermal comfort, aesthetic value, and attract beneficial insects, such as bees.

Frequency-Aware Feature Fusion for Dense Image Prediction.

Dense image prediction tasks demand features with strong category information and precise spatial boundary details at high resolution. To achieve this, modern hierarchical models often utilize feature fusion, directly adding upsampled coarse features from deep layers and high-resolution features from lower levels. In this paper, we observe rapid variations in fused feature values within objects, resulting in intra-category inconsistency due to disturbed high-frequency features. Additionally, blurred boundaries in fused features lack accurate high frequency, leading to boundary displacement. Building upon these observations, we propose Frequency-Aware Feature Fusion (FreqFusion), integrating an Adaptive Low-Pass Filter (ALPF) generator, an offset generator, and an Adaptive High-Pass Filter (AHPF) generator. The ALPF generator predicts spatially-variant low-pass filters to attenuate high-frequency components within objects, reducing intra-class inconsistency during upsampling. The offset generator refines large inconsistent features and thin boundaries by replacing inconsistent features with more consistent ones through resampling, while the AHPF generator enhances high-frequency detailed boundary information lost during downsampling. Comprehensive visualization and quantitative analysis demonstrate that FreqFusion effectively improves feature consistency and sharpens object boundaries. Extensive experiments across various dense prediction tasks confirm its effectiveness.

Transformer Dil-DenseUnet: An Advanced Architecture for Stroke Segmentation.

We propose a novel architecture, Transformer Dil-DenseUNet, designed to address the challenges of accurately segmenting stroke lesions in MRI images. Precise segmentation is essential for diagnosing and treating stroke patients, as it provides critical spatial insights into the affected brain regions and the extent of damage. Traditional manual segmentation is labor-intensive and error-prone, highlighting the need for automated solutions. Our Transformer Dil-DenseUNet combines DenseNet, dilated convolutions, and Transformer blocks, each contributing unique strengths to enhance segmentation accuracy. The DenseNet component captures fine-grained details and global features by leveraging dense connections, improving both precision and feature reuse. The dilated convolutional blocks, placed before each DenseNet module, expand the receptive field, capturing broader contextual information essential for accurate segmentation. Additionally, the Transformer blocks within our architecture address CNN limitations in capturing long-range dependencies by modeling complex spatial relationships through multi-head self-attention mechanisms. We assess our model's performance on the Ischemic Stroke Lesion Segmentation Challenge 2015 (SISS 2015) and ISLES 2022 datasets. In the testing phase, the model achieves a Dice coefficient of 0.80 ± 0.30 on SISS 2015 and 0.81 ± 0.33 on ISLES 2022, surpassing the current state-of-the-art results on these datasets.

Efficient retinal artery/vein classification with dense color-invariant feature learning

Efficient retinal artery/vein classification with dense color-invariant feature learning

Robust Multi-Class Classification for Real-Time Agricultural Applications Using Efficient and Adaptive Deep Learning

Plant diseases severely affect agricultural productivity, necessitating accurate and rapid detection methods. This research presents a robust, multi-class plant disease classification framework using adaptive deep learning. We utilize pre-trained convolutional neural networks (CNNs), specifically Xception, InceptionResNetV2, InceptionV3, ResNet50, and the proposed EfficientNetB3-based Adaptive Augmented Deep Learning (EfficientNetB3-AADL) model. Our approach leverages transfer learning combined with extensive data augmentation and trimming techniques to enhance model performance and mitigate overfitting. The EfficientNetB3-AADL architecture incorporates convolutional and max pooling layers, regularization strategies, and a dense feature learning layer, optimized to classify 52 disease categories from a publicly available leaf image dataset. The model’s performance is extensively evaluated using metrics such as accuracy, precision, recall, and F1 score. Notably, EfficientNetB3-AADL achieves superior accuracy over 98%, outperforming other CNN models. The proposed methodology highlights the efficacy of compound scaling and adaptive data augmentation in ensuring robust and efficient disease classification, suitable for real-time agricultural applications. This advancement supports sustainable farming by offering a scalable, computationally efficient solution for early and accurate disease detection in diverse crop species.

GKE-TUNet: Geometry-Knowledge Embedded TransUNet Model for Retinal Vessel Segmentation Considering Anatomical Topology.

Automated retinal vessel segmentation is crucial for computer-aided clinical diagnosis and retinopathy screening. However, deep learning faces challenges in extracting complex intertwined structures and subtle small vessels from densely vascularized regions. To address these issues, we propose a novel segmentation model, called Geometry-Knowledge Embedded TransUNet (GKE-TUNet), which incorporates explicit embedding of topological features of retinal vessel anatomy. In the proposed GKE-TUNet model, a skeleton extraction network is pre-trained to extract the anatomical topology of retinal vessels from refined segmentation labels. During vessel segmentation, the dense skeleton graph is sampled as a graph of key-points and connections and is incorporated into the skip connection layer of TransUNet. The graph vertices are used as node features and correspond to positions in the low-level feature maps. The graph attention network (GAT) is used as the graph convolution backbone network to capture the shape semantics of vessels and the interaction of key locations along the topological direction. Finally, the node features obtained by graph convolution are read out as a sparse feature map based on their corresponding spatial coordinates. To address the problem of sparse feature maps, we employ convolution operators to fuse sparse feature maps with low-level dense feature maps. This fusion is weighted and connected to deep feature maps. Experimental results on the DRIVE, CHASE-DB1, and STARE datasets demonstrate the competitiveness of our proposed method compared to existing ones.

Adversarial Unsupervised Domain Adaptation for 3D Semantic Segmentation with 2D Image Fusion of Dense Depth

AbstractUnsupervised domain adaptation (UDA) is increasingly used for 3D point cloud semantic segmentation tasks due to its ability to address the issue of missing labels for new domains. However, most existing unsupervised domain adaptation methods focus only on uni‐modal data and are rarely applied to multi‐modal data. Therefore, we propose a cross‐modal UDA on multi‐modal datasets that contain 3D point clouds and 2D images for 3D Semantic Segmentation. Specifically, we first propose a Dual discriminator‐based Domain Adaptation (Dd‐bDA) module to enhance the adaptability of different domains. Second, given that the robustness of depth information to domain shifts can provide more details for semantic segmentation, we further employ a Dense depth Feature Fusion (DdFF) module to extract image features with rich depth cues. We evaluate our model in four unsupervised domain adaptation scenarios, i.e., dataset‐to‐dataset (A2D2 → SemanticKITTI), Day‐to‐Night, country‐to‐country (USA → Singapore), and synthetic‐to‐real (VirtualKITTI → SemanticKITTI). In all settings, the experimental results achieve significant improvements and surpass state‐of‐the‐art models.

Mid-Summer observations of water column physical properties and circulation in the San Jorge Gulf (Patagonian Shelf)

In this study, we present detailed insights into the mid-summer density field and flow patterns within the San Jorge Gulf (Patagonian Shelf of Argentina). Utilizing unique data acquired from a towed undulating vehicle equipped with a Conductivity-Temperature-Depth (CTD) sensor and a hull-mounted acoustic Doppler current profiler (ADCP), we investigate the spatial distribution and temporal evolution of the density structure and associated currents. Our observations reveal the presence of a distinctive bottom dome-like structure comprised of dense, cold, and saline waters in the central basin of the gulf during mid-summer. Analysis of the flow dynamics indicates the presence of a near-geostrophic flow regime sustaining this dense water feature. Furthermore, our study highlights the significant role of ageostrophic velocities, primarily influenced by the modulation of pycnocline thickness by M2+M4 internal tides. These findings contribute to a deeper understanding of the oceanographic processes governing the mid-summer dynamics in the San Jorge Gulf, shedding light on the interaction between density structures and associated currents. Such insights are essential for advancing our knowledge of coastal ocean circulation and its implications for various ecological and environmental phenomena.

Efficient large-scale scene representation with a hybrid of high-resolution grid and plane features

Existing neural radiance fields (NeRF) methods for large-scale scene modeling require days of training using multiple GPUs, hindering their applications in scenarios with limited computing resources. Radiance field models based on explicit dense or hash grid features have been proposed for fast optimization, but their effectiveness has mainly been demonstrated in the context of object-scale scene representation. In this paper, we point out that the low feature resolution in explicit representation is the bottleneck for large-scale unbounded scene representation. To address this problem, we introduce a new and efficient hybrid feature representation for NeRF that fuses the 3D hash-grids and high-resolution 2D dense plane features. Compared with the dense-grid representation, the resolution of a dense 2D plane can be scaled up more efficiently. Based on this hybrid representation, we propose a fast optimization NeRF variant, called GP-NeRF, that achieves better rendering results while maintaining a compact model size. Extensive experiments on multiple large-scale unbounded scene datasets show that our model can converge in 1.5 h using a single GPU while achieving results comparable to or even better than the existing method that requires about one day’s training with 8 GPUs. Our code can be found at https://zyqz97.github.io/GP_NeRF/.

FCS-Net: Feather condition scoring of broilers based on dense feature fusion of RGB and thermal infrared images

Assessing the feather condition of broilers is crucial for monitoring the animal welfare status and detecting the occurrence of feather pecking activities. Currently, the feather condition of individual broilers is manually scored by trained experts. To provide a more objective and efficient tool for feather condition scoring, a novel deep learning-based model, named Feather Condition Scoring Network (FCS-Net), was proposed based on RGB and thermal infrared images. The FCS-Net model combined the ResNet18 architecture with the proposed Dense Feature Fusion (DFF) module, which can effectively learn the feature mapping relationship between RGB and thermal infrared images. Before inputting the images into the network, an image registration process was conducted to align the RGB and thermal infrared images. The results showed that the FCS-Net model had a good performance for feather condition scoring, with the Accuracy of 97.02%, the Precision of 96.99%, the Recall of 97.04%, the F1 of 97.01%, and the Inference speed of 15.34 fps. Compared to the ResNet18_RGB model, which only utilise RGB images, the FCS-Net model showed notable improvements in Accuracy by 4.02%, Precision by 3.90%, Recall by 4.08%, and F1 by 4.01%. Moreover, it was observed that the FCS-Net model focused more on the back region of the broilers through heatmap visualization. Furthermore, the algorithm was compared with six typical image recognition algorithms including VGG16, ResNet18, SE-ResNet18, DenseNet121, Mobilenet_V2, and Shufflenet_V2_x1_0, as well as the state-of-the-art (SOTA) feather condition assessment methods. The results showed that the FCS-Net model achieved better performance than the six algorithms and the SOTA feather condition assessment methods. This study provided a valuable reference for automated monitoring of feather condition scoring of broilers in smart farming.

Leveraging denoising diffusion probabilistic model to improve the multi-thickness CT segmentation

Organs-at-risk (OARs) segmentation in computed tomography (CT) is a fundamental step in the radiotherapy workflow, which has been prone as a time-consuming and labor-intensive task. Deep neural networks (DNNs) have gained significant popularity in the field of OAR segmentation tasks, achieving remarkable progress in clinical practice. Typically, OARs are distributed throughout different areas of the body and require varying thicknesses of CT scans for better diagnosis and segmentation in clinical. Most DNN-based segmentation focuses on single-thickness CT scans, limiting their applicability to varying thicknesses due to a lack of diverse thickness-related feature learning. While pre-training with the denoising diffusion probabilistic model (DDPM) offers an effective solution for dense feature learning, current works are constrained in addressing feature diversity, as exemplified by scenarios such as multi-thickness CT. To address the above challenges, this paper introduces a novel pre-training approach called DiffMT. This approach leverages the DDPM to extract valuable features from multi-thickness CT images. By transferring the pre-trained DDPM to the downstream segmentation for fine-tuning, the model gains proficiency in learning diverse multi-thickness CT features, leading to precise segmentation across varied thicknesses. We explore DiffMT’s feature learning capacity through experiments involving pre-trained models of varying sizes and different denoising thicknesses. Subsequently, thorough experiments comparing DDPM-based segmentation with other state-of-the-art (SOTA) CT segmentation methods, along with assessments on diverse OARs and modalities, empirically demonstrate that the proposed DiffMT method outperforms the control methods. The codes are available at https://github.com/ychengrong/DiffMT.

Radiogenomics based survival prediction of small-sample glioblastoma patients by multi-task DFFSP model.

In this paper, the multi-task dense-feature-fusion survival prediction (DFFSP) model is proposed to predict the three-year survival for glioblastoma (GBM) patients based on radiogenomics data. The contrast-enhanced T1-weighted (T1w) image, T2-weighted (T2w) image and copy number variation (CNV) is used as the input of the three branches of the DFFSP model. This model uses two image extraction modules consisting of residual blocks and one dense feature fusion module to make multi-scale fusion of T1w and T2w image features as backbone. Also, a gene feature extraction module is used to adaptively weight CNV fragments. Besides, a transfer learning module is introduced to solve the small sample problem and an image reconstruction module is adopted to make the model anatomy-aware under a multi-task framework. 256 sample pairs (T1w and corresponding T2w MRI slices) and 187 CNVs of 74 patients were used. The experimental results show that the proposed model can predict the three-year survival of GBM patients with the accuracy of 89.1 %, which is improved by 3.2 and 4.7 % compared with the model without genes and the model using last fusion strategy, respectively. This model could also classify the patients into high-risk and low-risk groups, which will effectively assist doctors in diagnosing GBM patients.

Fault detection method for transmission line components based on lightweight GMPPD-YOLO

Abstract This paper designs a lightweight high-precision transmission line component detection model, named grouped dense, monotonic self-regularized, and partial faster convolution, pruning, and distillation optimized—you only look once (GMPPD-YOLO), in transmission line inspection. It addresses the issue of low detection accuracy of target detection algorithms due to the complex background, large differences in target shape, location, texture, etc, as well as diversified and smaller defects in insulator and vibration hammer images taken by unmanned aerial vehicles from multiple angles. To enhance the model’s feature extraction capabilities in complex backgrounds and across different scales, the grouped dense C3 dense feature extraction module was designed, enabling the model to more effectively handle diverse defect forms. Simultaneously, the monotonic self-regularized pyramid pooling–fast (MSPPF) module is proposed to enhance the model’s capability to process multi-scale information. Additionally, the partial-faster C3 feature awareness module is designed to improve feature fusion performance, enhancing the model’s ability to perceive features at different scales. Finally, channel pruning was used to reduce redundant parameters, and knowledge distillation was employed to compensate for the accuracy loss caused by pruning. This approach further compressed the model size while ensuring its detection performance. The experimental results demonstrate that compared to the original YOLOv5s algorithm, the proposed GMPPD-YOLO algorithm achieves a reduction in parameters by 68.4%, a decrease in Giga floating-point operations per second by 58.2%, and a reduction in the model size by 66.4%, while achieving an increase in precision by 1%, mAP50 by 1.1%, and mAP95 by 0.4%. This confirms the significant potential of the GMPPD-YOLO algorithm for deployment in real-time drone-based power transmission line inspections.

Dense Feature Pyramid Deep Completion Network

Most current point cloud super-resolution reconstruction requires huge calculations and has low accuracy when facing large outdoor scenes; a Dense Feature Pyramid Network (DenseFPNet) is proposed for the feature-level fusion of images with low-resolution point clouds to generate higher-resolution point clouds, which can be utilized to solve the problem of the super-resolution reconstruction of 3D point clouds by turning it into a 2D depth map complementation problem, which can reduce the time and complexity of obtaining high-resolution point clouds only by LiDAR. The network first utilizes an image-guided feature extraction network based on RGBD-DenseNet as an encoder to extract multi-scale features, followed by an upsampling block as a decoder to gradually recover the size and details of the feature map. Additionally, the network connects the corresponding layers of the encoder and decoder through pyramid connections. Finally, experiments are conducted on the KITTI deep complementation dataset, and the network performs well in various metrics compared to other networks. It improves the RMSE by 17.71%, 16.60%, 7.11%, and 4.68% compared to the CSPD, Spade-RGBsD, Sparse-to-Dense, and GAENET.

Automatic identification of breech face impressions based on deep local features

Breech face impressions are an essential type of physical evidence in forensic investigations. However, their surface morphology is complex and varies based on the machining method used on the gun’s breech face, making traditional handcrafted local feature-based methods exhibit high false rates and are unsuitable for striated impressions. We proposed a deep local feature-based method for firearm identification utilizing Detector-Free Local Feature Matching with Transformers (LoFTR). This method removes the module of feature point detection and directly utilizes self and cross-attention layers in the Transformer to transform the convolved coarse-level feature maps into a series of dense feature descriptors. Subsequently, matches with high confidence scores are filtered based on the score matrix calculated from the dense descriptors. Finally, the screened initial matches are refined into the convolved fine-level features, and a correlation-based approach is used to obtain the exact location of the match. Validation tests were conducted using three authoritative sets of the breech face impressions datasets provided by the National Institute of Standards and Technology (NIST). The validation results show that, compared with the traditional handcrafted local-feature based methods, the proposed method in this paper yields a lower identification error rate. Notably, the method can not only deal with granular impressions, but can also be applied to the striated impressions. The results indicate that the method proposed in this paper can be utilized for comparative analysis of breech face impressions, and provide a new automatic identification method for forensic investigations.

Brain tumor classification for combining the advantages of multilayer dense net-based feature extraction and hyper-parameters tuned attentive dual residual generative adversarial network classifier using wild horse optimization.

In this manuscript, attentive dual residual generative adversarial network optimized using wild horse optimization algorithm for brain tumor detection (ADRGAN-WHOA-BTD) is proposed. Here, the input imageries are gathered using BraTS, RemBRANDT, and Figshare datasets. Initially, the images are preprocessed to increase the quality of images and eliminate the unwanted noises. The preprocessing is performed with dual-tree complex wavelet transform (DTCWT). The image features like geodesic data and texture features like contrasts, energy, correlations, homogeneity, and entropy are extracted using multilayer dense net methods. Then, the extracted images are given to attentive dual residual generative adversarial network (ADRGAN) classifier for classifying the brain imageries. The ADRGAN weight parameters are tuned based on wild horse optimization algorithm (WHOA). The proposed method is executed in MATLAB. For the BraTS dataset, the ADRGAN-WHOA-BTD method achieved accuracy, sensitivity, specificity, F-measure, precision, and error rates of 99.85%, 99.82%, 98.92%, 99.76%, 99.45%, and 0.15%, respectively. Then, the proposed technique demonstrated a runtime of 13 s, significantly outperforming existing methods.

Approaching Angstrom-Scale Resolution in Lithography Using Low-Molecular-Mass Resists (

Resists that enable high-throughput and high-resolution patterning are essential in driving the semiconductor technology forward. The ultimate patterning performance of a resist in lithography is limited because of the trade-off between resolution, line-width roughness, and sensitivity; improving one or two of these parameters typically leads to a loss in the third. As the patterned feature sizes approach angstrom scale, the trade-off between these three metrics becomes increasingly hard to resolve and calls for a fundamental rethinking of the resist chemistry. Low-molecular-mass monodispersed metal-containing resists of high atom economy can provide not only very high resolution but also very low line-width roughness without sacrificing sensitivity. Here we describe a modular metal-containing resist platform (molecular mass <500 Da) where a molecular resist consists of just two components: a metal and a radical initiator bonded to it. This simple system not only is amenable to high-resolution electron beam lithography (EBL) and extreme ultraviolet lithography (EUVL) but also unites them mechanistically, giving a consolidated perspective of molecular and chemical processes happening during exposure. Irradiation of the resist leads to the production of secondary electrons that generate radicals in the initiator bonded to metal. This brings about an intramolecular rearrangement and causes solubility switch in the exposed resist. We demonstrate record 1.9-2.0 nm isolated patterns and 7 nm half-pitch dense line-space features over a large area using EBL. With EUVL, 12 nm half-pitch line-space features are shown at a dose of 68 mJ/cm2. In both of these patterning techniques, the line-width roughness was found to be ≤2 nm, a record low value for any resist platform, also leading to a low-performance trade-off metric, Z factor, of 0.6 × 10-8 mJ·nm3. With the ultimate resolution limited by instrumental factors, potential patterning at the level of a unit cell can be envisaged, making low-molecular-mass resists best poised for angstrom-scale lithography.

Billion-scale pre-trained knowledge graph model for conversational chatbot

Conversational Chatbot is a critical technology for managing customer service on E-commerce platforms. However, most Chatbots struggle with data scarcity and sparsity. To tackle these issues, a novel two-step (pretrain and fine-tune) click-through-rate prediction model called knowledge-enhanced deep cross network (K-DCN) is proposed. First, a billion-scale conversation knowledge graph is constructed, inspired by real-world application scenarios. Then, the conversation knowledge graph is pretrained to capture the semantic and structure information respectively. Our K-DCN leverages novel user-state and dialogue-interaction representations from the pre-trained conversation knowledge graph, integrating them with the conventional dense and sparse feature representations within the Deep Cross Network framework. This integration is designed to perform item ranking in recommendation. From the experiments, our proposed K-DCN significantly outperforms the baseline models, and the model has been deployed in real world application.