A5076092831- Spectroscopy and Chemometric Analyses
- Advanced Image and Video Retrieval Techniques
- Video Surveillance and Tracking Methods
- Face recognition and analysis
- Advanced Image Processing Techniques
- Advanced Chemical Sensor Technologies
- Advanced Neural Network Applications
- Generative Adversarial Networks and Image Synthesis
- Image Retrieval and Classification Techniques
- Music and Audio Processing
- Smart Agriculture and AI
- Advanced Image Fusion Techniques
- Autonomous Vehicle Technology and Safety
- Multimodal Machine Learning Applications
- Phytochemicals and Antioxidant Activities
- Video Analysis and Summarization
- Human Pose and Action Recognition
- Complex Network Analysis Techniques
- Text and Document Classification Technologies
- Face and Expression Recognition
- Essential Oils and Antimicrobial Activity
- Topic Modeling
- Robotics and Sensor-Based Localization
- Advanced Measurement and Detection Methods
- Identification and Quantification in Food
Anhui Normal University
2019-2025
State Key Laboratory of Transducer Technology
2021
Shanghai Institute of Microsystem and Information Technology
2021
Peking University
2021
University of Nottingham Ningbo China
2016-2020
First People's Hospital of Chongqing
2020
Nanjing Agricultural University
2016-2019
Shenzhen University
2018-2019
National Administration of Surveying, Mapping and Geoinformation of China
2019
University of Nottingham
2018
We present a novel method for constructing Variational Autoencoder (VAE). Instead of using pixel-by-pixel loss, we enforce deep feature consistency between the input and output VAE, which ensures VAE's to preserve spatial correlation characteristics input, thus leading have more natural visual appearance better perceptual quality. Based on recent learning works such as style transfer, employ pre-trained convolutional neural network (CNN) use its hidden features define loss VAE training....
This paper presents an advanced urban traffic density estimation solution using the latest deep learning techniques to intelligently process ultrahigh-resolution videos taken from unmanned aerial vehicle (UAV). We first capture nearly hour-long video at five busy road intersections of a modern megacity by flying UAV during rush hours. then randomly sampled over 17 K 512×512 pixel image patches frames and manually annotated 64 vehicles form dataset for this paper, which will also be made...
Lexical analysis is believed to be a crucial step towards natural language understanding and has been widely studied. Recent years, end-to-end lexical models with recurrent neural networks have gained increasing attention. In this report, we introduce deep Bi-GRU-CRF network that jointly word segmentation, part-of-speech tagging named entity recognition tasks. We trained the model using several massive corpus pre-tagged by our best Chinese tool, together small, yet high-quality human...
ABSTRACT Effective conservation of endangered species necessitates not only the preservation core habitats but also enhancement landscape connectivity. As a critically Crocodylia, Chinese alligator ( Alligator sinensis ) strongly relies on fragmented wetland habitat lower area Yangtze River. The integrity its needs evaluating, and connectivity restoring plan designing. In this study, we estimated suitability in River using Maxent model. Then, potential ecological corridors between each...
Despite 40 years of conservation the critically endangered Chinese alligator ( Alligator sinensis ), genomic underpinnings its status remained uncharted. Genome sequencing data 244 individuals uncovered relatively low overall diversity/heterozygosity and long runs homozygosity, with captive populations exhibiting higher heterozygosity smaller inbreeding coefficients compared to wild individuals. The decreased level in population demonstrates contribution large breeding population. estimated...
Near-infrared (NIR) spectroscopy was applied for the determination of total soluble solid contents (SSC) single Ruby Seedless grape berries using both benchtop Fourier transform (VECTOR 22/N) and portable grating scanning (SupNIR-1500) spectrometers in this study. The results showed that best SSC prediction obtained by VECTOR 22/N range 12,000 to 4000 cm−1 (833–2500 nm) with coefficient (Rp2) 0.918, root mean squares error (RMSEP) 0.758% based on least support vector machine (LS-SVM)....
The design of deep graph models still remains to be investigated and the crucial part is how explore exploit knowledge from different hops neighbors in an efficient way. In this paper, we propose a novel RNN-like neural network architecture by incorporating AdaBoost into computation network; proposed convolutional called AdaGCN~(Adaboosting Graph Convolutional Network) has ability efficiently extract high-order current nodes then integrates Adaboost Different other networks that directly...
We have developed a learning-based image transformation framework and successfully applied it to three common operations: downscaling, decolorization, high dynamic range tone mapping. use convolutional neural network (CNN) as non-linear mapping function transform an input desired output. A separate CNN trained for very large classification task is used feature extractor construct the training loss of CNN. Unlike similar applications in related literature such super-resolution, none problems...
The double-yolked (DY) egg is quite popular in some Asian countries because it considered as a sign of good luck, however, the double yolk one reasons why these eggs fail to hatch. usage automatic methods for identifying DY can increase efficiency poultry industry by decreasing loss during incubation or improving sale proceeds. In this study, two duck identification were developed using computer vision technology. Transmittance images and single-yolked (SY) acquired CCD camera identify them...
This paper presents a novel indoor topological localization method based on mobile phone videos. Conventional methods suffer from dynamic environmental changes and scene ambiguity. The proposed Visual Landmark Sequence-based Indoor Localization (VLSIL) is capable of addressing problems by taking steady objects as landmarks. Unlike many feature or appearance matching-based methods, our utilizes highly abstracted landmark sematic information to represent locations thus invariant illumination...
For a city to be livable and walkable is the ultimate goal of future cities. However, conflicts among pedestrians, vehicles, cyclists at traffic intersections are becoming severe in high-density urban transportation areas, especially China. Correspondingly, transit time prolonged, pedestrian safety endangered. Simulating movements complex necessary optimize organization. We propose an unmanned aerial vehicle (UAV)-based method for tracking simulating intersections. Specifically,...
Photoplethysmography (PPG) is an important signal which contains much physiological information like heart rate and cardiovascular health etc. However, PPG signals are easily corrupted by motion artifacts body movements during their recordings, may lead to poor quality. In order accurately extract information, it necessary ensure high quality in these applications. Although there several existed methods get the quality, those algorithms complex accuracies not very high. Thus, this work...
Space-time adaptive processing (STAP) is an effective tool for detecting a moving target in spaceborne or airborne radar systems. Statistical-based STAP methods generally need sufficient statistically independent and identically distributed (IID) training data to estimate the clutter characteristics. However, most actual scenarios appear only locally stationary lack IID data. In this paper, by exploiting intrinsic sparsity of distribution angle-Doppler domain, new algorithm called SR-STAP...
Vehicle behavior recognition is an attractive research field which useful for many computer vision and intelligent traffic analysis tasks. This paper presents all-in-one framework moving vehicles based on the latest deep learning techniques. Unlike traditional methods rely low-resolution videos captured by road cameras, we capture 4K ( 3840 × 2178 ) at a busy intersection of modern megacity flying unmanned aerial vehicle (UAV) during rush hours. We then manually annotate locations types...