Recently, network representation learning has been widely used to mine and analyze characteristics, it is also applied blockchain, but most of the embedding methods in blockchain ignore heterogeneity network, so difficult accurately describe characteristics transaction. As smart society evolves, Ethereum makes contracts reality, while transaction appearing on platform scarce; thus, there an urgent need from contract transfer. In this article, we propose a heterogeneous method implicit...

Multi-view clustering can partition data samples into their categories by learning a consensus representation in unsupervised way and has received more attention recent years. However, most existing deep methods learn or view-specific representations from multiple views via view-wise aggregation way, where they ignore structure relationship of all samples. In this paper, we propose novel multi-view network to address these problems, called Global Cross-view Feature Aggregation for Multi-View...

Combining color (RGB) images with thermal can facilitate semantic segmentation of poorly lit urban scenes. However, for RGB-thermal (RGB-T) segmentation, most existing models address cross-modal feature fusion by focusing only on exploring the samples while neglecting connections between different samples. Additionally, although importance boundary, binary, and information is considered in decoding process, differences complementarities morphological features are usually neglected. In this...

Crowd density estimation has gained significant research interest owing to its potential in various industries and social applications. Therefore, this paper proposes a multistyle joint-perception network based on knowledge distillation-trained student (MJPNet-S*) for drone-based red–green–blue, thermal/depth (RGB-T/D) crowd tasks. To provide superior accuracy efficiency, novel trimodal working module effectively combines the modalities facilitate comprehensive extraction utilization. A...

Owing to the development of convolutional neural networks (CNNs), detection defects on rail surfaces has significantly improved. Although existing methods achieve good results, they incur huge computational and parameter costs associated with CNNs. The usual approach this problem is design lightweight models that meet needs real-world applications; however, performance often compromised. To address aforementioned problems, we designed a dual semantic approximation network via knowledge...

Urban scene parsing is the core of intelligent transportation system, and RGB–thermal urban has recently attracted increasing research interest in field computer vision. However, most existing approaches fail to perform good boundary extraction for prediction maps cannot fully use high-level features. In addition, these methods simply fuse features from RGB thermal modalities but are unable obtain comprehensive fused To address problems, an edge-aware guidance fusion network (EGFNet) was...

Multiview clustering (MVC) has gained significant attention as it enables the partitioning of samples into their respective categories through unsupervised learning. However, there are a few issues follows: 1) many existing deep methods use same latent features to achieve conflict objectives, namely, reconstruction and view consistency. The objective aims preserve view-specific for each individual view, while view-consistency strives obtain common across all views; 2) some embedded (DEC)...

The rapid progression of convolutional neural networks (CNNs) has significantly improved indoor scene parsing, transforming the fields robotics, autonomous navigation, augmented reality, and surveillance. Currently, societal demand is propelling these technologies toward integration into mobile smart device applications. However, processing capabilities devices cannot support comprehensive system requirements CNNs, which poses a challenge for several deep-learning One promising solution to...

In this paper, a new high-resolution approach called fourth-order cumulants-based Toeplitz matrices reconstruction (FOC-TMR) method, is presented for two-dimensional (2-D) direction-of-arrival (DOA) estimation of incident narrowband coherent signals. The angle problem addressed by arranging the cumulants elements received signals from two parallel uniform linear arrays (ULAs) to matrices. Gaussian noise cases, it shown that ranks equal number incoming waves and are independent their...

Deep learning has become a popular method for studying the semantic segmentation of high-resolution remote sensing images (HRRSIs). Existing methods have adopted convolutional neural networks to achieve better accuracy HRRSIs, and success these models often depends on model complexity parameter quantity. However, deployment equipment with limited resources is significant challenge. To solve this problem, lightweight student network framework—a graph attention guidance (GAGNet) knowledge...

Camouflaged object detection (COD) is an important yet challenging task, with great application values in industrial defect detection, medical care, etc. The challenges mainly come from the high intrinsic similarities between target objects and background. In this paper, inspired by biological studies that consists of two steps, i.e., search identification, we propose a novel framework, named DCNet, for accurate COD. DCNet explores candidate extra object-related edges through constraints...

Applying computer vision techniques to rail surface defect detection (RSDD) is crucial for preventing catastrophic accidents. However, challenges such as complex backgrounds and irregular shapes persist. Previous methods have focused on extracting salient object information from a pixel perspective, thereby neglecting valuable high- low-frequency image information, which can better capture global structural information. In this study, we design pixel-aware frequency conversion network...

Multiview data, characterized by rich features, are crucial in many machine learning applications. However, effectively extracting intraview features and integrating interview information present significant challenges multiview (MVL). Traditional deep network-based approaches often involve multiple layers to derive latent. In these methods, the of different classes typically implicitly embedded rather than systematically organized. This lack structure makes it challenging explicitly map...

Two-dimensional (2-D) direction-of-arrival (DOA) estimation method using three-parallel uniform linear arrays (ULAs) is proposed in this letter. The 2-D DOA problem addressed by making full use of elements the ULAs. Furthermore, algorithm has better angle performance practical mobile elevation angles between 70° and 90° can automatically pair estimated azimuth with lower complexity. Simulation results demonstrate effectiveness algorithm.

Nowadays, the standard dynamic range (SDR) image acquired at a fixed exposure exposes weakness in portraying fine-grained details of real scenes. The high (HDR) and other types SDR images generated by multiexposure fusion techniques provide us new choices for scene representation. To display on screens, an HDR must be tone-mapped to one. Since different tone-mapping/fusion algorithms produce with varying visual quality levels, it naturally desires evaluation model comparison. This article...

Remote sensing image segmentation plays an important role in many industrial-grade processing applications. However, the problem of uncertainty caused by intraclass heterogeneity and interclass blurring is prevalent high-resolution remote images. Moreover, complexity information images leads to a large amount background around objects. To solve this problem, new fuzzy convolutional neural network proposed paper. This resolves ambiguity feature introducing neighbourhood module deep learning...

Owing to the expansion in processing of industrial information through advances machine learning, demand for accurate crowd counting various applications is increasing. We propose a multimodality cross-guided compensation coordination network (MC <inline-formula xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink"> <tex-math notation="LaTeX">$^{3}$</tex-math> </inline-formula> Net) red–green–blue and thermal (RGB-T) counting. The includes modules...

Deep learning techniques have largely solved the problem of rail surface defect detection (SDD), however, two aspects yet to be addressed. In most existing approaches, red–green–blue and depth (RGB-D) streams are indiscriminately fused across modalities, ignoring fact that RGB images produce different feature qualities in scenes. Additionally, their focus on performance, previous studies overlooked models several parameters, resulting unrealistic practical applications. To address these...