A5100687859- Generative Adversarial Networks and Image Synthesis
- Advanced Image Processing Techniques
- Neural dynamics and brain function
- Lung Cancer Treatments and Mutations
- Neurobiology and Insect Physiology Research
- Cancer Genomics and Diagnostics
- Visual perception and processing mechanisms
- Image Enhancement Techniques
- Simulation and Modeling Applications
- Advanced Vision and Imaging
- Advanced Neural Network Applications
- Image Processing Techniques and Applications
- Computer Graphics and Visualization Techniques
- Advanced Measurement and Detection Methods
- Video Analysis and Summarization
- Vehicle Dynamics and Control Systems
- Video Surveillance and Tracking Methods
- Human Pose and Action Recognition
- Esophageal Cancer Research and Treatment
- Advanced Algorithms and Applications
- Virus-based gene therapy research
- Cancer Research and Treatments
- Advanced Optical Sensing Technologies
- Industrial Technology and Control Systems
- EEG and Brain-Computer Interfaces
Henan Cancer Hospital
2017-2025
Zhengzhou University
2015-2025
Zhejiang University
2021-2024
Japan Advanced Institute of Science and Technology
2024
Xi'an Jiaotong University
2024
Shandong Freshwater Fisheries Research Institute
2024
Shanghai University of Electric Power
2023
Zhejiang University of Science and Technology
2019-2023
Ludong University
2023
ORCID
2021
Although existing image inpainting approaches have been able to produce visually realistic and semantically correct results, they only one result for each masked input. In order multiple diverse reasonable solutions, we present Unsupervised Cross-space Translation Generative Adversarial Network (called UCTGAN) which mainly consists of three network modules: conditional encoder module, manifold projection module generation module. The the are combined learn one-to-one mapping between two...
Image style transfer is an underdetermined problem, where a large number of solutions can satisfy the same constraint (the content and style). Although there have been some efforts to improve diversity by introducing alternative loss, they restricted generalization, limited poor scalability. In this paper, we tackle these limitations propose simple yet effective method for diversified arbitrary transfer. The key idea our operation called deep feature perturbation (DFP), which uses orthogonal...
Artistic style transfer is an image editing task that aims at repainting everyday photographs with learned artistic styles. Existing methods learn styles from either a single example or collection of artworks. Accordingly, the stylization results are inferior in visual quality limited controllability. To tackle this problem, we propose novel Dual Style-Learning Style Transfer (DualAST) framework to simultaneously both holistic artist-style (from artworks) and specific artwork-style image):...
Recent studies have shown remarkable success in universal style transfer which transfers arbitrary visual styles to content images. However, existing approaches suffer from the aesthetic-unrealistic problem that introduces disharmonious patterns and evident artifacts, making results easy spot real paintings. To address this limitation, we propose AesUST, a novel Aesthetic-enhanced Universal Style Transfer approach can generate aesthetically more realistic pleasing for styles. Specifically,...
Content and style (C-S) disentanglement is a fundamental problem critical challenge of transfer. Existing approaches based on explicit definitions (e.g., Gram matrix) or implicit learning GANs) are neither interpretable nor easy to control, resulting in entangled representations less satisfying results. In this paper, we propose new C-S disentangled framework for transfer without using previous assumptions. The key insight explicitly extract the content information implicitly learn...
Arbitrary style transfer (AST) transfers arbitrary artistic styles onto content images. Despite the recent rapid progress, existing AST methods are either incapable or too slow to run at ultra-resolutions (e.g., 4K) with limited resources, which heavily hinders their further applications. In this paper, we tackle dilemma by learning a straightforward and lightweight model, dubbed MicroAST. The key insight is completely abandon use of cumbersome pre-trained Deep Convolutional Neural Networks...
Parkinson's disease (PD) is a prevalent neurodegenerative disorder characterized by dopaminergic neuron death in the substantia nigra, leading to motor dysfunction. Autophagy dysregulation has been implicated PD pathogenesis. This study explores role of miR-214-3p PD, focusing on its impact autophagy and viability. Using vitro vivo models, we demonstrate that inhibits promotes apoptosis. Behavioral assessments molecular analyses reveal exacerbation symptoms upon overexpression. Furthermore,...
The wavelet transform as an important multiresolution analysis tool has already been commonly applied to texture and classification. Nevertheless, it ignores the structural information while capturing spectral of image at different scales. In this paper, we propose a classification approach with linear regression model based on transform. This method is motivated by observation that there exists distinctive correlation between sample images, belonging same kind texture, frequency regions...
This paper presents a new adversarial training framework for image inpainting with segmentation confusion (SCAT) and contrastive learning. SCAT plays an game between generator network, which provides pixel-level local signals can adapt to images free-form holes. By combining standard global training, the exhibits following three advantages simultaneously: (1) consistency of repaired image, (2) fine texture details (3) flexibility handling Moreover, we propose textural semantic learning...
Abstract The monitor of sea fogs become more important with the rapid development marine activities. Remote sensing through laser is an effective tool for monitoring fogs, but still challengeable large distance. We demonstrated a Long-distance Lidar fog superconducting nanowire single-photon detector (SNSPD), which extended ranging area to 180-km diameter area. system, was verified by using benchmark distance measurement known island, applied Mie scattering weather prediction system. echo...
Recently proposed LaMa [25] introduce Fast Fourier Convolution (FFC) [4] into image inpainting. FFC empowers the fully convolutional network to have a global receptive field in its early layers, and ability produce robust repeating texture. However, has difficulty generating clear sharp complex content. In this paper, we analyze fundamental flaws of using inpainting, which are 1) spectrum shifting, 2) unexpected spatial activation, 3) limited frequency field. Such make FFC-based inpainting...
Cancer patients are highly susceptible to infections due their immunocompromised state from both the malignancy and intensive treatments. Accurate timely identification of causative pathogens is crucial for effective management treatment. Targeted next-generation sequencing (tNGS) has become an important tool in clinical infectious disease diagnosis because its broad microbial detection range acceptable cost. However, there currently a lack systematic research evaluate diagnostic value this...
Accurate target detection in natural environments is an important function of the visual systems vertebrates and has a direct impact on animal survival environmental adaptation. Existing studies have shown that mammalian prefrontal cortex plays role detection. However, mechanisms brain regions similar to other species, such as avian nidopallium caudolaterale, not been well studied. Here, we selected pigeons, known for their excellent ability, model studied dynamic changes caudolaterale...
Image style transfer aims to the styles of artworks onto arbitrary photographs create novel artistic images. Although is inherently an underdetermined problem, existing approaches usually assume a deterministic solution, thus failing capture full distribution possible outputs. To address this limitation, we propose Diverse Style Transfer (DIST) framework which achieves significant diversity by enforcing invertible cross-space mapping. Specifically, consists three branches: disentanglement...
Owing to the increased use of urban rail transit, flow passengers on metro platforms tends increase sharply during peak periods. Monitoring passenger in such areas is important for security-related reasons. In this paper, order solve problem platform detection, we propose a CNN (convolutional neural network)-based network called MP (metro platform)-CNN accurately count people platforms. The proposed method composed three major components: group convolutional networks used front end extract...
Temperature evolution of the train brake disc during high-speed braking was investigated using in situ experiments, theoretical analysis, and finite element modeling. The experimental results show that temperature distribution on friction surface experienced formation a hot ring first, then expansion duration ring. Alternative spot cold zone were observed surface, which is attributed to local contact couple heterogeneous heat dissipation condition disc. corresponding maximum increased...
Detecting moving objects in a video sequence is an important problem many vision-based applications. In particular, detecting when the camera difficult problem. this study, we propose symmetric method for presence of dynamic background. First, background compensation used to detect proposed region motion. Next, order accurately locate objects, convolutional neural network-based called YOLOv3-SOD all image, which lightweight and specifically designed small objects. Finally, are determined by...
Abstract Esophageal squamous cell carcinoma (ESCC) is highly heterogeneous. Our understanding of full molecular and immune landscape ESCC remains limited, hindering the development personalised therapeutic strategies. To address this, we perform genomic-transcriptomic characterizations AI-aided histopathological image analysis 120 Chinese patients. Here show that can be categorized into differentiated, metabolic, immunogenic stemness subtypes based on bulk single-cell RNA-seq, each...
A novel method of all-wheel braking force allocation during a braking-in-turn maneuver is proposed for vehicles with the brake-by-wire (BBW) system. The concept stability priority applied in distributing to each wheel, which means that performance has over demand maneuver. Based on this principle, I-curves maneuvers respect three parameters are developed using graphics single-track vehicle model, defining between front and rear axles. Then, distributed axles allocated inner outer wheels...