A5086501723- Advanced Neural Network Applications
- Video Surveillance and Tracking Methods
- Autonomous Vehicle Technology and Safety
- Robotics and Sensor-Based Localization
- Advanced Vision and Imaging
- Spine and Intervertebral Disc Pathology
- Optical measurement and interference techniques
- Remote Sensing and LiDAR Applications
- Image Enhancement Techniques
- Human Pose and Action Recognition
- Spinal Fractures and Fixation Techniques
- Medical Imaging and Analysis
- Image Processing Techniques and Applications
- Advanced Image and Video Retrieval Techniques
- Image and Signal Denoising Methods
- Robotic Path Planning Algorithms
- Musculoskeletal pain and rehabilitation
- Anomaly Detection Techniques and Applications
- Domain Adaptation and Few-Shot Learning
- UAV Applications and Optimization
- Vehicle Dynamics and Control Systems
- Advanced Image Processing Techniques
- Advanced Image Fusion Techniques
- Advanced Optical Sensing Technologies
- Advanced Measurement and Detection Methods
Tongji University
2020-2025
Beijing Jishuitan Hospital
2018-2024
Peking University
2012-2024
Minzu University of China
2024
Capital Medical University
2022-2024
Hong Kong Polytechnic University
2023
Beijing Tian Tan Hospital
2020-2023
Beijing Institute of Technology
2023
Chinese Academy of Medical Sciences & Peking Union Medical College
2022
University of Jinan
2012-2022
Locating 3D objects from a single RGB image via Perspective-n-Points (PnP) is long-standing problem in computer vision. Driven by end-to-end deep learning, recent studies suggest interpreting PnP as differentiable layer, so that 2D-3D point correspondences can be partly learned backpropagating the gradient w.r.t. object pose. Yet, learning entire set of unrestricted points scratch fails to converge with existing approaches, since deterministic pose inherently non-differentiable. In this...
Object localization in 3D space is a challenging aspect monocular object detection. Recent advances 6DoF pose estimation have shown that predicting dense 2D-3D correspondence maps between image and model then estimating via Perspective-n-Point (PnP) algorithm can achieve remarkable accuracy. Yet these methods rely on training with ground truth of geometry, which difficult to acquire real outdoor scenes. To address this issue, we propose MonoRUn, novel detection framework learns...
3D-aware image synthesis encompasses a variety of tasks, such as scene generation and novel view from images. Despite numerous task-specific methods, developing comprehensive model remains challenging. In this paper, we present SSDNeRF, unified approach that employs an expressive diffusion to learn generalizable prior neural radiance fields (NeRF) multi-view images diverse objects. Previous studies have used two-stage approaches rely on pretrained NeRFs real data train models. contrast,...
As a variant of rapidly exploring random tree (RRT), RRT* is an important improvement sampling-based algorithms. Although it can provide feasible planning solution with higher quality, more resources on optimization are required, resulting in very slow convergence rate, which cannot satisfy the real-time requirements most autonomous systems. In this paper, we propose novel approach collaboration double-tree structure to separate extension and procedure. our algorithm, original RRT employed...
Motion planning is one of the most significant technologies for autonomous driving. To make motion models able to learn from environment and deal with emergency situations, a new framework called as "parallel planning" proposed in this paper. In order generate sufficient various training samples, artificial traffic scenes are firstly constructed based on knowledge reality. A deep model which combines convolutional neural network (CNN) Long Short-Term Memory module (LSTM) developed decisions...
Deep convolutional neural networks have been applied by automobile industries, Internet giants, and academic institutes to boost autonomous driving technologies; while progress has witnessed in environmental perception tasks, such as object detection driver state recognition, the scene-centric understanding identification still remain a virgin land. This mainly encompasses two key issues: 1) lack of shared large datasets with comprehensively annotated road scene information 2) difficulty...
Visual tracking of multiple objects is an essential component for a perception system in autonomous driving vehicles. One the favorable approaches tracking-by-detection paradigm, which links current detection hypotheses to previously estimated object trajectories (also known as tracks) by searching appearance or motion similarities between them. As this search operation usually based on very limited spatial temporal locality, association can fail cases noise long-term occlusion. In paper, we...
Video object detection has drawn great attention recently. The Vision Meets Drone Object Detection in Challenge 2019 (VisDrone-VID2019) is held to advance the state-of-the-art video for videos captured by drones. Specifically, there are 13 teams participating challenge. We also report results of 6 detectors on collected dataset. A short description provided appendix each detector. present analysis and discussion challenge results. Both dataset publicly available at website: http://www.aiskyeye.com/.
The rapid advances of transportation infrastructure have led to a dramatic increase in the demand for smart systems capable monitoring traffic and street safety. Fundamental these applications are community-based evaluation platform benchmark object detection multi-object tracking. To this end, we organize AVSS2017 Challenge on Advanced Traffic Monitoring, conjunction with International Workshop Street Surveillance Safety Security (IWT4S), evaluate state-of-the-art tracking algorithms...
Locating 3D objects from a single RGB image via Perspective-n-Point (PnP) is long-standing problem in computer vision. Driven by end-to-end deep learning, recent studies suggest interpreting PnP as differentiable layer, allowing for partial learning of 2D-3D point correspondences backpropagating the gradients pose loss. Yet, entire scratch highly challenging, particularly ambiguous solutions, where globally optimal theoretically non-differentiable w.r.t. points. In this paper, we propose...
Tissue engineering has become a rapidly developing field of research because the increased demand from regenerative medicine, and hydrogels are promising tissue scaffold their three-dimensional structures. In this study, we constructed novel gelatin methacrylate (GelMA) modified with histidine Zn2+ (GelMA-His-Zn(II)), which possessed fascinating antibacterial properties tunable mechanical formation functionalized dual network covalent crosslinking metal coordination bonds. The introduction...
Imitation learning for the end-to-end autonomous driving has drawn renewed attention from academic communities. Current methods either only use images as input, which will yield ambiguities when a vehicle approaches an intersection, or additional command information to navigate but inefficiently. Focusing on making automatically drive along given path, we propose new and effective navigation called subgoal angle does not require human participation is calculated by current position of...
Generally, the extended Kalman filter (EKF) is used for sensor fusion in a land vehicle navigation system. However, defects of first-order linearization nonlinear model EKF can introduce large estimated errors, and may lead to sub-optimal performance. In order yield higher accuracy navigation, this paper, novel particle (PF) proposed sampling importance resampling (SIR-PF) applied address measurement it shows better performances when compared with EKF. The basic theories application general...
To improve the efficiency of three-dimensional (3-D) LIDAR mapping, multiple robots cooperation has been considered in mapping large areas. In order to merge local maps acquired by independent robots, it is crucial identify common areas among maps. this article, we propose a novel 3-D cooperative approach for connected and automated vehicles that use only data cooperatively create globally consistent robots. Assuming each individual vehicle able its own map, process divided into two main...
The changes in the microenvironment of degenerative intervertebral discs cause oxidative stress injury and excessive apoptosis disc endogenous stem cells. purpose this study was to explore possible mechanism protective effect melatonin on NPMSCs induced by H2O2.The Cell Counting Kit-8 assay used evaluate cytotoxicity hydrogen peroxide effects melatonin. ROS content detected 2'7'-dichlorofluorescin diacetate (DCFH-DA). Mitochondrial membrane potential (MMP) JC-1assay. Transferase mediated...
Driver distraction behavior recognition is currently a significant study area that involves analyzing and identifying various movements, actions, patterns exhibited by drivers while operating vehicles. This field has received considerable attention due to its potential enhance driving safety through driver monitoring tasks, widely implemented in advanced assistance systems autonomous As result, extensive efforts have been made utilize different sensor modalities algorithms understand...
Respiratory motion-induced vertebral movements can adversely impact intraoperative spine surgery, resulting in inaccurate positional information of the target region and unexpected damage during operation. In this paper, we propose a novel deep learning architecture for respiratory motion prediction, which adapt to different patients. The proposed method utilizes an LSTM-AE with attention mechanism network that be trained using few-shot datasets To ensure real-time performance, dimension...
Abstract We investigated topographic roughness for the northern hemisphere (>45°N) of Mercury using high‐resolution topography data acquired by Laser Altimeter (MLA) on board MErcury Surface, Space ENvironment, GEochemistry, and Ranging (MESSENGER) spacecraft. Our results show that there are distinct differences in bidirectional slope root‐mean‐square (RMS) height among smooth plains (SP), intercrater (ICP), heavily cratered terrain (HCT), ratios RMS three geologic units both about...