A5019199537- Human Pose and Action Recognition
- Video Surveillance and Tracking Methods
- Advanced Vision and Imaging
- Metaheuristic Optimization Algorithms Research
- Gait Recognition and Analysis
- Anomaly Detection Techniques and Applications
- Advanced Multi-Objective Optimization Algorithms
- Advanced Optimization Algorithms Research
- Human Motion and Animation
- Traffic Prediction and Management Techniques
- Plant biochemistry and biosynthesis
- Evolutionary Algorithms and Applications
- Neural Networks and Applications
- Advanced Image Processing Techniques
- Robotic Path Planning Algorithms
- Time Series Analysis and Forecasting
- Ginkgo biloba and Cashew Applications
- Speech Recognition and Synthesis
- Robotics and Sensor-Based Localization
- Computer Graphics and Visualization Techniques
- Emotion and Mood Recognition
- Advanced Neural Network Applications
- Advanced Image and Video Retrieval Techniques
Nanjing University of Posts and Telecommunications
2016-2025
Sichuan University
2006
For the autonomous driving system, accurately recognizing actions of different roles in traffic scene is prerequisite for realizing this kind human-vehicle information interaction. In paper, we propose a complete framework based on 3D human pose estimation to recognize road. The main objects recognized include police, cyclists, and some passersby need. We perform action recognition dynamic adaptive graph convolutional network, which can realize pose. addition module, have optimized both...
Abstract An acaricidal substance extracted from the external seed coat of Ginkgo biloba L. was identified by UV (ultraviolet), IR (infrared), EI‐MS (electron impact ion source mass spectrometry), 1 H NMR (nuclear magnetic resonance) and 13 C as 6‐[( Z )‐10‐heptadecenyl]‐2‐hydroxybenzoic acid (compound 1). Laboratory bioassay on citrus red mite, Panonychus citri (Mcg), showed that compound possessed following properties. (i) Powerful contact toxicity with an LC 50 5.2 mg litre −1 after 24 h...
Estimating human pose and shape without markers is a challenging problem. This study proposes multiple‐view markerless motion capture framework. Firstly, multi‐view camera system built for capturing real‐time images of moving humans on multiple views. Secondly, by employing the OpenPose method, authors calculate robust 3D key points from 2D body, which are estimated images. And dense point cloud reconstructed Thirdly, they propose novel SMPL‐based method to represent fitting SMPL model...
Visual tracking has become a hot research topic in the field of computer vision at present. The process can be converted into dynamic optimization problem. In this article, an image fitness metric which named Structural Similarity Index (SSIM) is used to estimate object location experiments. addition, DE (differential evolution), population based stochastic meta-heuristic algorithm adopted for According characteristics visual tracking, article initialized particles by Gaussian distribution...
Image registration is effect by finding the best spatial transformation between reference image and floating in terms of optimizing a similarity metric. Mutual information, as an effective method, reliable criterion for registration. In this paper, we propose improved differential evolution algorithm with replacement strategy to maximize mutual information multi-temporal Experimental results show that our new achieves more precise robust when compared several other classical optimization methods.
The artificial bee colony (ABC) algorithm invented by Karaboga is a relative new nature inspired heuristic for optimization problem. It has been proved to be competitive with some conventional algorithms. This paper proposes an adaptive unified (auABC) which employs single equation unifying multiple strategies into one expression. As we all know, ABC good at exploration but poor exploitation due the insufficiency in its solution search and one-dimensional strategy leads slower convergence...