A5101568187- Advanced Neural Network Applications
- Advanced Image and Video Retrieval Techniques
- Cloud Computing and Resource Management
- Domain Adaptation and Few-Shot Learning
- Advanced Vision and Imaging
- IoT and Edge/Fog Computing
- Visual Attention and Saliency Detection
- Photonic and Optical Devices
- Human Pose and Action Recognition
- Quantum Computing Algorithms and Architecture
- Multimodal Machine Learning Applications
- Plasmonic and Surface Plasmon Research
- Advanced Image Fusion Techniques
- Machine Learning and ELM
- Complex Network Analysis Techniques
- Video Surveillance and Tracking Methods
- Quantum Information and Cryptography
- Distributed and Parallel Computing Systems
- Advanced Algorithms and Applications
- Anomaly Detection Techniques and Applications
- Orbital Angular Momentum in Optics
- Neural Networks and Applications
- Adversarial Robustness in Machine Learning
- Remote-Sensing Image Classification
- Generative Adversarial Networks and Image Synthesis
China General Nuclear Power Corporation (China)
2025
Zhengzhou University
2025
National Supercomputing Center in Wuxi
2025
Inspur (China)
2016-2025
China Southern Power Grid (China)
2024
China Electronic Information Industry Development
2018-2024
Jining First People's Hospital
2024
Shenzhen University
2018
Beihang University
2013-2014
Image Inpainting has recently become an important research problem due to the rise of generative image synthesis models. While many solutions have been proposed for this problem, it is challenging establish a testbed different possible types inpainting masks e.g., completion mask, expand thick brushes etc. Most shine on object removal or texture synthesis, while semantic generation still difficult achieve. To address these issues, we introduce first general Challenge. The target develop that...
Solving the depth estimation problem in a 360° image space, which has holistic scene perception, become trend recent years. However, common images is prone to geometric distortion. Therefore, this study proposes new method, CAPDepth, address geometric-distortion of monocular estimation. We reduce tangential projections by an optimized content-aware projection (CAP) and embedding module capture more features for global consistency. Additionally, we adopt index map de-blocking scheme improve...
Distributed deep learning has become an essential technique for accelerating learning, but its performance is often influenced by the heterogeneous computing nodes and communication networks within distributed platform. Due to high costs of practical deployment running it almost impossible researchers optimize training strategy tasks in real-world environments. In this paper, we propose HeterSim, a simulator specifically designed platforms. HeterSim enables flexible configuration node...
The moisture content is closely related to the shear creep deformation behavior of soft rock, and linear rock can be described by classical Nishihara model. However, its accuracy in describing accelerated nonlinear characteristics effects still needs improved. innovation this paper propose an improved model that describe whole process shale with different content. uses a strain-triggered sticky pot proposes damage factor reflect effect on rock. relationship between exponential function,...
We demonstrate the optical trapping of single dielectric nanoparticles in a microfluidic chamber using coupled T-shaped copper plasmonic nanoantenna for studying light–matter interaction. The is composed two identical elements separated by 50 nm gap and each element designed with nanoblocks. Our inherits three different advantages compared to previous nanoantennas, which are usually made gold. First, very promising material its similar properties as Second, comparably cheap, compatible...
Existing deep learning based remote sensing change detection (RSCD) methods only rely on binary ground-truth to guide the network while neglecting useful semantic guidance. As a result, can be readily misled by irrelevant category changes, leading degraded performance and slow convergence of model. To this end, we propose novel segment anything model (SAM) guided framework, termed as SAM-CD, which mines rich knowledge from SAM for RSCD. Specifically, first employ transformer encoder extract...
Representing features at multiple scales is significant for person re-identification (Re-ID). Most existing methods learn the multi-scale by stacking streams and convolutions without considering cooperation of a granular level. However, most are more discriminative only when they integrate other as contextual information. We termed that multi-scale. In this paper, we proposed novel architecture, namely network (CMSNet), learning common representations simultaneously. The building block...
Abstract In recent years, the field of dielectric-plasmonic photonics has made remarkable strides, leading to successful development various technologies. The realization sophisticated optical circuits on a single platform become increasingly viable. Here we propose and investigate hybrid dielectric waveguide integrated with plasmonics. This comprises copper nanowire situated in close proximity silicon V-groove channel, separated by nanoscale gap. configuration is particularly advantageous,...
Multimodal medical image fusion is a method of integrating information from multiple formats. Its aim to provide useful and accurate for doctors. Multi-channel pulse coupled neural network (m-PCNN) recently proposed m
Hyperparameter optimization is a challenging problem in developing deep neural networks. Decision of transfer layers and trainable major task for design the convolutional networks (CNN). Conventional CNN models are usually manually designed based on intuition. In this paper, genetic algorithm applied to select model. The filter criterion constructed by accuracy counts layers. results show that method competent task. system will converge with precision 97% classification Cats Dogs datasets,...
Co-salient object detection (CoSOD) aims to discover and segment foreground targets in a group of images with the same semantic category. Existing mainstream approaches often employ convolutional neural networks (CNNs) learn semantic-invariant features from images. Despite demonstrated success, there exist two limitations: 1) The CNNs introduce inductive bias locality that are difficult model long-range dependency, limiting their feature representation capability. 2) Their models lack...
Simulation of quantum computing on supercomputers is a significant research topic, which plays vital role in algorithm verification, error-tolerant and other applications. Tensor-network contraction based density matrix an important single-amplitude simulation strategy, but it hard to execute the distributed systems. In this paper, we studied problem detail, propose scheme cutting edges undirected graphs. This cuts graphs with large tree width obtain many subgraphs small width, these...
Visual Question Answering (VQA) serves as a proxy for evaluating the scene understanding of an intelligent agent by answering questions about images. Most VQA benchmarks to date are focused on those that can be answered through visual content in scene, such simple counting, attributes, and even little challenging require extra encyclopedic knowledge. However, humans have remarkable capacity reason dynamic interaction which is beyond literal image has not been investigated so far. In this...
Shor algorithm is a key milestone in the quantum computing history. It has greatly promoted development of since it supposed to crack RSA cryptosystems. Simulation technology an important strategy research, but software simulation can only simulate final result rather than behavior and hardware structure. This paper proposes for algorithm. The closer execution process computation simulation. helps understand how computer works more concrete, guide design specifically. According workflow...
Research in the field of photonic integrated circuits (PICs) is taking a boost, especially because its compatibility with modern complementary metal-oxide semiconductor fabrication technology. Silicon-on-insulator slot waveguides are burgeoning platform for sophisticated on-chip integration applications and have been extensively leveraged PICs. Here structural optimization parametric analysis waveguide geometry optical enhancement nanoscale confinement C band presented. Theoretical...
Three‐photon microscopy excited at the 1700‐nm window enables deep‐tissue penetration. However, refractive indices of commonly used immersion oils, and resultant pulse broadening are not known, preventing imaging optimization. Here, we demonstrate detailed characterization index, distortion for excitation pulses this oils. On physical side, uncover that absorption, rather than material dispersion, is main cause distortion. application comparative three‐photon results indicate 1600‐nm yields...