Few-shot learning presents a challenge that classifier must quickly adapt to new classes do not appear in the training set, given only few labeled examples of each class. This paper proposes position-aware relation network (PARN) learn more flexible and robust metric ability for few-shot learning. Relation networks (RNs), kind architectures relational reasoning, can acquire deep images by just being designed as simple convolutional neural (CNN)[23]. However, due inherent local connectivity...

Emerging Metaverse applications demand reliable, accurate, and photorealistic reproductions of human hands to perform sophisticated operations as if in the physical world. While real hand represents one most intricate coordination between bones, muscle, tendon, skin, state-of-the-art techniques unanimously focus on modeling only skeleton hand. In this paper, we present NIMBLE, a novel parametric model that includes missing key components, bringing 3D new level realism. We first annotate...

Epicardial adipose tissue (EAT) contribute to atrial fibrillation (AF). We sought explore the role of FAI (fat attention index), volume, fat radiomic profile (FRP) from peri-coronary artery (PCAT) on coronary computed tomography angiography (CCTA) in determining presence AF and differentiating types. This study enrolled 300 patients underwent CCTA retrospectively divided them into (n = 137) non-AF groups 163). The imaging parameters FAI, volume FRP were excavated measured after PCAT...

This work aims to explore an accurate and effective method for recognizing dance movement features, providing precise personalized guidance sports teaching. First, a human skeletal graph is constructed. A convolutional network (GCN) employed extract features from the nodes (joints) edges (bone connections) in structure, capturing both spatial relationships temporal dynamics between joints. The GCN generates motion representations by aggregating of each node its neighboring nodes. recognition...

The creation of 3D contents still remains one the most crucial problems for emerging applications such as printing and Augmented Reality. In Reality, how to create virtual that seamlessly overlay with real environment is a key problem human-computer interaction many subsequent applications. this paper, we present sketch-based interactive tool, which term emph{SweepCanvas}, rapid exploratory modeling on top an RGB-D image. Our aim offer end-users simple yet efficient way quickly models We...

This paper presents a fully automatic framework for extracting editable 3D objects directly from single photograph. Unlike previous methods which recover either depth maps, point clouds, or mesh surfaces, we aim to with semantic parts and can be edited. We base our work on the assumption that most human-made are constituted by these well represented generalized primitives. Our makes an attempt towards recovering two types of primitive-shaped objects, namely, cuboids cylinders. To this end,...

In networks, a link prediction task aims at learning potential relations between nodes to predict unknown linkage states. At present, most methods are used process static networks. These cannot produce good results for dynamic However, networks in the real world, vertices and links of these change over time. Dynamic (DLP) has attracted more attention as it can better mimic evolution nature Inspired by successful applications generative adversarial network generating fake images, which...

Hand modeling is critical for immersive VR/AR, action understanding, or human healthcare. Existing parametric models account only hand shape, pose, texture, without the anatomical attributes like bone, which essential realistic biomechanics analysis. In this paper, we present PIANO, first bone model of hands from MRI data. Our PIANO biologically correct, simple to animate, and differentiable, achieving more anatomically precise inner kinematic structure in a data-driven manner than...

With the rapid development of deep learning model in computer vision area, it has outperformed most traditional machine algorithms. Since tire factories pay much attention to defects detection tires based on x-ray image, lots image methods are brought up. However, there still challenges accuracy. This paper put forward a novel and modified Faster R-CNN conduct detection. Some proper processing is done before extracting features detecting then adjusting feature extractor, proposal generator...

The Voxel-Based Nearest Neighbor method (VNN) is the most commonly used three-dimensional ultrasound (US) image reconstruction algorithm for spine imaging on scoliotic patients. neural radiance field (NeRF) a newly developed in computer vision area, which utilizes weighted multilayer perceptron to rebuild volumetric density and color of 3D scene. objectives this study are apply NeRF US data evaluate spinal curvature measurement from reconstructed results. conventional VNN were reconstruct...

This paper presents a secure decentralized clustering algorithm for wireless ad-hoc sensor networks. The operates without centralized controller, asynchronously, and does not require that the location of sensors be known priori. Based on cluster-based topology, hierarchical communication protocols dynamic quarantine strategies are introduced to defend against spam attacks, since this type attacks can exhaust energy nodes will shorten lifetime network drastically. By adjusting threshold...

Image retargeting methods aim to minimize the perceptual loss while changing sizes and aspect ratios of images. Since optimal for different images are generally not same, image quality assessment (IRQA) becomes a meaningful task. This paper proposes content-aware method using foreground global measurement achieve better performance. In our proposed method, first divided into two categories according object detection result, then corresponding measurements designed them. For those with...

Acute myocardial infarction, a critical condition in cardiovascular disease, is associated with high degree of morbidity and mortality. Early identification the risk acute infarction (AMI) can provide treatment to patients as early possible effectively reduce death. In this study, we constructed an AMI mortality prediction model based on MIMIC-IV database explored relationship between relevant data characteristics effects. We included into study sample, used SMOTE algorithm correct sample...

Aimed at the path coverage problem for wireless multimedia sensor networks, MTPCA (Moving Target based Path Coverage-enhancing Algorithm) algorithm is proposed. This introduces improved probabilistic model into virtual potential field directional network coverage. In order to correct centroid position of sector area covered by node in traditional model, this puts forward concept impact factor obtain better effect. Furthermore, MTPCA, nodes acquire real-time information target point through...

Download This Paper Open PDF in Browser Add to My Library Share: Permalink Using these links will ensure access this page indefinitely Copy URL DOI

Abstract Aiming at the problems that positioning accuracy of laser Simultaneous Localization and Mapping (SLAM) is affected by misclassification close feature points incorrect matching in environment with redundant features, a 3D SLAM algorithm based on graph optimization was proposed. Firstly, to reduce extraction error problem, distance threshold used filter point cloud improve classification, enhanced reliability optoelectronic measurement data. Secondly, consistency matching, rough...