A5086628680- Face and Expression Recognition
- Video Surveillance and Tracking Methods
- Image Retrieval and Classification Techniques
- Advanced Image and Video Retrieval Techniques
- Vehicle License Plate Recognition
- Advanced Neural Network Applications
- Human Pose and Action Recognition
- Hand Gesture Recognition Systems
- Face recognition and analysis
- Handwritten Text Recognition Techniques
- Image Processing Techniques and Applications
- Cell Image Analysis Techniques
- Plasma and Flow Control in Aerodynamics
- Digital Imaging for Blood Diseases
- Remote-Sensing Image Classification
- Multimodal Machine Learning Applications
- Anomaly Detection Techniques and Applications
- Legume Nitrogen Fixing Symbiosis
- Sleep and Work-Related Fatigue
- AI in cancer detection
- Plant pathogens and resistance mechanisms
- Plasma Diagnostics and Applications
- Image and Object Detection Techniques
- Neural Networks and Applications
- Soybean genetics and cultivation
Second Affiliated Hospital of Nanchang University
2021-2025
Nanchang University
2021-2025
Ningbo University of Technology
2021-2024
Chinese Academy of Sciences
2024
University of Chinese Academy of Sciences
2024
Institute of Zoology
2024
Peking University
2023
Hiroshima University
2023
Hubei University of Technology
2019-2022
Harbin Institute of Technology
2016-2021
Current state-of-the-art image-text matching methods implicitly align the visual-semantic fragments, like regions in images and words sentences, adopt cross-attention mechanism to discover fine-grained cross-modal semantic correspondence. However, may bring redundant or irrelevant region-word alignments, degenerating retrieval accuracy limiting efficiency. Although many researchers have made progress mining meaningful alignments thus improving accuracy, problem of poor efficiency remains...
Driver fatigue and inattention have long been recognised as the main contributing factors in traffic accidents. This study presents a novel system which applies convolutional neural network (CNN) to automatically learn predict pre‐defined driving postures. The idea is monitor driver hand position with discriminative information extracted safe/unsafe posture. In comparison previous approaches, CNNs can features directly from raw images. authors' works, CNN model was first pre‐trained by an...
Vehicle-type recognition based on images is a challenging task. This paper comparatively studied two feature extraction methods for image description, i.e., the Gabor wavelet transform and Pyramid Histogram of Oriented Gradients (PHOG). The has been widely adopted to extract features various vision tasks. PHOG superiority in its description more discriminating information. A highly reliable classification scheme was proposed by cascade classifier ensembles with reject option accommodate...
Classification of medical images is an important issue in computer-assisted diagnosis. In this paper, a classification scheme based on one-class kernel principle component analysis (KPCA) model ensemble has been proposed for the images. The consists KPCA models trained using different image features from each class, and product combining rule was used to produce confidence scores assigning class. effectiveness verified breast cancer biopsy dataset 3D optical coherence tomography (OCT)...
With the massive deployment of distributed video surveillance systems, automatic detection abnormal events in streams has become an urgent need. An event can be considered as a deviation from regular scene; however, distribution normal and is severely imbalanced, since do not frequently occur. To make use large number videos scenes, we propose semi-supervised learning scheme, which only uses data that contains ordinary scenes. The proposed model two-stream structure composed appearance...
Detection and recognition of traffic sign, including various road signs text, play an important role in autonomous driving, mapping/navigation safety. In this paper, we proposed a sign detection system by applying deep convolutional neural network (CNN), which demonstrates high performance with regard to rate accuracy. Compared other published methods are usually limited predefined set signs, our is more comprehensive as target includes digits, English letters Chinese characters. The based...
Traffic safety is a severe problem around the world. Many road accidents are normally related with driver's unsafe driving behavior, e.g. eating while driving. In this work, we propose vision-based solution to recognize behavior based on convolutional neural networks. Specifically, given an image, skin-like regions extracted by Gaussian Mixture Model, which passed deep networks model, namely R*CNN, generate action labels. The able provide abundant semantic information sufficient...
The RNA-guided endonucleases of the CRISPR-Cas9 system, including most widely used Cas9 from Streptococcus pyogenes (SpCas9), are becoming a robust genome editing tool in model organisms and hold immense promise for therapeutic applications. Many strategies have been employed to overcome limitations caused by SpCas9's off-target effects its stringent requirement protospacer adjacent motif (PAM) sequence. However, structural mechanisms underlying these remain undefined. Here, we present...
In preclinical and phase I II clinical studies, 2'-deoxy-2'-β-fluoro-4'-azidocytidine (FNC) displays a potent long-lasting inhibition of HIV-1 infection. To investigate its mechanism action, we compared it with the well-documented lamivudine (3TC). Pharmacokinetic studies revealed that intracellular retention FNC triphosphate in peripheral blood mononuclear cells was markedly longer than 3TC triphosphate. selectively enters is retained HIV target cells, where exerts prevention addition to...
Contextual information plays an important role in visual recognition. This is especially true for action recognition as contextual information, such the objects a person interacts with and scene which performed, inseparable from predefined class. Meanwhile, attention mechanism of humans shows remarkable capability compared existing computer vision system discovering information. Inspired by this, we applied soft adding two extra branches original VGG16 model one to apply scene-level whilst...
Recognition of traffic signs is vary important in many applications such as self-driving car/driverless car, mapping and surveillance. Recently, deep learning models demonstrated prominent representation capacity, achieved outstanding performance sign recognition. In this paper, we propose a recognition system by applying convolutional neural network (CNN). comparison with previous methods which usually use CNN feature extractor multi-layer perception (MLP) classifier, proposed max pooling...
Gait as a biometric feature that can be measured remotely without physical contact and proximal sensing has attract significant attention. This paper proposes to use con-volutional neural networks (ConvNets) multi-task learning model(MLT) identify human gait predict multiple attributes simultaneously. In comparison previous approaches, two novelty in our convolutional approach summarised (i)using ConvNets learn rich features from the training set is more generic requires minimal domain...
Acinetobacter baumannii bacteremia caused by pandrug-resistant strains poses a major challenge in intensive care units, necessitating novel therapeutic approaches. Phage-derived depolymerases offer promising adjunct to conventional antibiotics. However, studies on A. phage have been limited non-mammalian models. This study investigates the efficacy, safety, and potential mechanisms of action DPO-HL, both as monotherapy combination with polymyxin B, murine model bacteremia. DPO-HL was...
Cross-linked enzyme crystals (CLECs) of subtilisin exhibit excellent activity in aqueous and various organic solvents. This catalyst is more stable than the native both mixed aqueous/organic solutions. Subtilisin-CLEC was shown to be a versatile catalyst. It used for syntheses peptides peptidomimetics, mild hydrolysis amino acid peptide amides, enantio- regioselective reactions, transesterifications.
Detection of traffic signs plays an important role in autonomous driving, surveillance and safety. Previous research Traffic Sign (TSD) generally focused on which are over the roads, road surface have not been discussed. In this paper, we propose a sign detection system by applying convolutional neural network (CNN). The proposed consists two main stages: 1) hybrid region proposal method to hypothesize locations taking into account complementary information color edge; 2) feature extraction,...