Fusing a low spatial resolution hyperspectral image with high multispectral has become popular for generating (HR-HSI). Most methods assume that the degradation information from to is known in and spectral domains. Conversely, this often limited or unavailable practice, restricting their performance. Furthermore, existing fusion still face problem of insufficient exploration cross-interaction between domains HR-HSI, leaving scope further improvement. This paper proposes an unsupervised...

Shape representation for 3-D models is an important topic in computer vision, multimedia analysis, and graphics. Recent multiview-based methods demonstrate promising performance shape recognition retrieval. However, most ignore the correlations of multiple views or suffer from high computional cost. In this paper, we propose a novel network architecture Our combines convolutional neural networks (CNNs) with long short-term memory (LSTM) to exploit correlative information views....

To address the high computational and memory cost in 3-D volumetric convolutional neural networks (CNNs), we propose an approach to train binary CNNs for object recognition. Our method is specifically designed data, which it transforms inputs weights convolutional/fully connected layers values, can potentially accelerate by efficient bitwise operations. Two loss calculation methods are solve accuracy decrease problem when last layer binarized. Four obtained from their corresponding...

Despite the advent in 3D hand pose estimation, current methods predominantly focus on single-image reconstruction camera frame, overlooking world-space motion of hands. Such limitation prohibits their direct use egocentric video settings, where hands and are continuously motion. In this work, we propose HaWoR, a high-fidelity method for world coordinates from videos. We to decouple task by reconstructing space estimating trajectory coordinate system. To achieve precise an adaptive SLAM...

Multi-Source Domain Adaptation (MSDA), which dedicates to transfer the knowledge learned from multiple source domains an unlabeled target domain, has drawn increasing attention in research community. By assuming that and share consistent key feature representations identical label space, existing studies on MSDA typically utilize entire union set of features both obtain map align for each category domain. However, default setting may neglect issue "partialness", i.e., 1) a part contained not...

A thermal camera captures the temperature distribution of a scene as image. In images, facial appearances different people under lighting conditions are similar. This is because generally constant and not affected by condition. similarity in face advantageous for detection. To detect faces cascade classifiers with Haar-like features used. However, there few studies exploring local detection images. this paper, we introduce two approaches relying on First, create new feature types extending...

Deep generative models often perform poorly in real-world applications due to the heterogeneity of natural data sets. Heterogeneity arises from containing different types features (categorical, ordinal, continuous, etc.) and same type having marginal distributions. We propose an extension variational autoencoders (VAEs) called VAEM handle such heterogeneous data. is a deep model that trained two stage manner first provides more uniform representation second stage, thereby sidestepping...

Many real-life decision-making situations allow further relevant information to be acquired at a specific cost, for example, in assessing the health status of patient we may decide take additional measurements such as diagnostic tests or imaging scans before making final assessment. Acquiring more enables better decision making, but costly. How can trade off desire make good decisions by acquiring with cost performing that acquisition? To this end, propose principled framework, named EDDI...

With the help of continuous optimizations in hardware and software, smartphones can now capture vivid, detailed macro pictures as well high-resolution videos. How-ever, taking photos/videos a low-light environment with would still result underexposed bad-quality due to their physical limitations — small sensor size, compact lenses, lack specific software. A variety enhancement techniques have been proposed, but effectiveness is limited by high complexity computational resources smartphones....

Tracking multiple tiny objects is highly challenging due to their weak appearance and limited features. Existing multi-object tracking algorithms generally focus on singlemodality scenes, overlook the complementary characteristics of captured by remote sensors. To enhance performance integrating information from sources, we propose a novel framework called HGT-Track (Heterogeneous Graph Transformer based Multi-Tiny-Object Tracking). Specifically, first employ Transformer-based encoder embed...

Flow field super-resolution (FFSR) aims at recovering high-resolution turbulent velocity fields from low-resolution flow fields. Existing FFSR methods mainly process the in natural image patterns, while critical and distinct fluid visual properties are rarely considered. This negligence would cause significant domain gap between images to severely hamper accurate perception of flows, thereby undermining performance a wrong pattern. To tackle this dilemma, we rethink task with properties,...

This work provides a novel methodology to non-destructively detect far-sided defect of sandwich structures by local resonance-based acoustic activation. Sandwich panel specimens with 10 mm thick balsa wood core and 25 Nomex honeycomb are investigated, artificial facesheet-core disbond size 30 × is introduced the specimens. Finite element modeling constructed investigate resonance frequency region. Experimental apparatus defect, where vibration activation measurement performed at near-sided...

Common Services researched by Service Science have four major provision modes: web page mode, local client service mode and cloud mode. Currently the most common used is Web And now provided every Provider access discovery strategy still based on keyword search's UDDI. returned to users often can't be called. This paper will raise a new which combines with Search Engine technology Semantic technology. engine provides search capability lexical analysis grammar while matching scoring semantic...

In view of terrain classification the autonomous multi-legged walking robots, two synthetic methods for classification, Simple Linear Iterative Clustering based Support Vector Machine (SLIC-SVM) and SegNet (SLIC-SegNet), are proposed. SLIC-SVM is proposed to solve problem that SVM can only output a single label fails identify mixed terrain. The SLIC-SegNet single-input multi-output model derived improve applicability classifier. Since results high quality legged robot use hard gain, obtains...

Deep neural networks have achieved great progress in 3D scene understanding. However, recent methods mainly focused on objects with canonical orientations contrast random postures reality. In this letter, we propose a hierarchical network, named Local Frame Network (LFNet), based the local rotation invariant coordinate frame for robust point cloud analysis. The patches different orientated are transformed into an identical distribution frame, and coordinates taken as input features to...

Recent RGBD-based models for saliency detection have attracted research attention. The depth clues such as boundary clues, surface normal, shape attribute, etc., contribute to the identification of salient objects with complicated scenarios. However, most RGBD networks require multi-modalities from input side and feed them separately through a two-stream design, which inevitably results in extra costs on sensors computation. To tackle these inconveniences, we present this paper novel fusion...

A new type of polymeric rheology modifier was synthesized by suspension polymerization, and the effect on rheological properties oil-based drilling fluids investigated. The results indicated that obtained polymer had good capacity improvement shearing force under high temperature pressure conditions. Moreover, can improve stability greatly. As a result, is for fluids, it optimize fluid system with properties, static ability cuttings environmental protection function. It play an essential...