A5103086654- Image Processing Techniques and Applications
- Advanced Vision and Imaging
- Random lasers and scattering media
- Advanced Optical Imaging Technologies
- Optical measurement and interference techniques
- Remote Sensing and LiDAR Applications
- Machine Learning in Healthcare
- Fault Detection and Control Systems
- Robotics and Sensor-Based Localization
- Forensic Anthropology and Bioarchaeology Studies
- Face Recognition and Perception
- Advanced Image and Video Retrieval Techniques
- Nanofluid Flow and Heat Transfer
- Neural Networks and Reservoir Computing
- Forensic and Genetic Research
- Advanced Optical Sensing Technologies
- Optical Imaging and Spectroscopy Techniques
- Interpreting and Communication in Healthcare
- Modular Robots and Swarm Intelligence
- Heat Transfer and Optimization
- Artificial Intelligence in Healthcare
- Innovation Diffusion and Forecasting
- Robot Manipulation and Learning
- Industrial Vision Systems and Defect Detection
- Educational Management and Quality
Southeast University
2021-2024
University Town of Shenzhen
2023
Tsinghua University
2023
Xi'an Jiaotong University
2018-2021
Columbia University
2021
Beijing Jingshida Electromechanical Equipment Research Institute
2020
University of Chicago
2019
Beijing University of Chemical Technology
2014
Extracting roads from satellite imagery is a promising approach to update the dynamic changes of road networks efficiently and timely. However, it challenging due occlusions caused by other objects complex traffic environment, pixel-based methods often generate fragmented fail predict topological correctness. In this paper, motivated shapes connections in graph network, we propose connectivity attention network (CoANet) jointly learn segmentation pair-wise dependencies. Since strip...
Even though ghost imaging (GI), an unconventional method, has received increased attention by researchers during the last decades, speed is still not satisfactory. Once data-acquisition method and system parameters are determined, only processing potential to accelerate image-processing significantly. However, both basic correlation compressed sensing algorithm, which often used for imaging, have their own problems. To overcome these challenges, a novel deep learning proposed in this paper....
Much of the recent efforts on salient object detection (SOD) have been devoted to producing accurate saliency maps without being aware their instance labels. To this end, we propose a new pipeline for end-to-end segmentation (SIS) that predicts class-agnostic mask each detected instance. better use rich feature hierarchies in deep networks and enhance side predictions, regularized dense connections, which attentively promote informative features suppress non-informative ones from all...
True Digital Orthophoto Maps (TDOMs) have high geometric accuracy and rich image characteristics, making them essential geographic data for national economic social development. Complex terrain artificial structures, automatic distortion elimination occluded area recovery in TDOM generation pose significant challenges. Hence, the need further improvements both mapping automation is highlighted. In this paper, we present an approach generating a based on Neural Radiance Field (NeRF) without...
Three-dimensional (3D) imaging under the condition of weak light and low signal-to-noise ratio is a challenging task. In this paper, 3D scheme based on time-correlated single-photon counting technology proposed demonstrated. The scheme, which composed pulsed laser, scanning mirror, detectors, module, employs for LiDAR (Light Detection Ranging). Aided by range-gated technology, experiments show that can image object when decreased to −13 dB improve structural similarity index results 10...
In order to solve the problem of small obstacles avoidance in ultra-low altitude flying unmanned aerial vehicles (UAVs), a depth estimation algorithm based on monocular visual for tower is proposed by combining deep learning and traditional methods. With advantage convolutional neural networks extract salient features images, lightweight multi-level residual convolution network obstacle segmentation proposed. The cross-entropy loss function optimized adding weight background area. A method...
Based on optical correlations, ghost imaging is usually reconstructed by computer algorithm from the acquired data. We here proposed an alternatively high contrast naked-eye scheme which avoids processing. Instead, uses a photoelectric feedback loop to realize multiplication process of traditional imaging. Meanwhile, it exploits vision persistence effect implement integral and generate negative images observed naked eyes. To imaging, special pattern-scanning architecture low-speed...
The EX16+22Y system is a polymerase chain reaction (PCR)-based amplification kit that enables typing of 15 autosomal short tandem repeat (STR) loci (i.e., D3S1358, D13S317, D7S820, D16S539, TPOX, TH01, D2S1338, CSF1PO, D19S433, vWA, D18S51, D21S11, D8S1179, D5S818, and FGA) 22 widely used Y chromosome STR (Y-STR) (DYS391, DYS527a/b, DYS635, DYS458, DYS456, DYS385a/b, DYS438, DYS448, DYS437, DYS19, DYS576, DYS533, DYS393, DYS389I/II, DYS439, DYS392, Y_GATA_H4, DYS390, DYS481) which contains...
Leveraging large amounts of longitudinal electronic health records (EHR) data has shown great potentials for predicting progression various diseases. In this work, we proposed a deep learning algorithm based on customized Long Short-Term Memory (LSTM) model with built-in temporal sequence mechanism, named as Order-Aware Medical SeQuence Learning (OA-MedSQL). The can predict clinical outcomes from EHR sequences competitive accuracies, meanwhile automatically distills patient medical journeys...