A5103027131- Atomic and Molecular Physics
- Reinforcement Learning in Robotics
- Robotics and Sensor-Based Localization
- Video Analysis and Summarization
- Cancer-related molecular mechanisms research
- Smart Agriculture and AI
- Rock Mechanics and Modeling
- Video Surveillance and Tracking Methods
- Computational Geometry and Mesh Generation
- Human Pose and Action Recognition
- Mass Spectrometry Techniques and Applications
- Data Stream Mining Techniques
- Ion-surface interactions and analysis
- IoT and Edge/Fog Computing
- Advanced Image and Video Retrieval Techniques
- Mobile Crowdsensing and Crowdsourcing
- Advanced Chemical Physics Studies
- Remote Sensing in Agriculture
- Cloud Computing and Resource Management
- Geological Modeling and Analysis
- Remote Sensing and Land Use
- Laser-Matter Interactions and Applications
- Neural Networks and Reservoir Computing
- Domain Adaptation and Few-Shot Learning
- 3D Surveying and Cultural Heritage
Jiangnan University
2024
Hunan University
2024
Shenzhen University
2023
Guizhou University
2023
Institute of Modern Physics
2016-2018
Chinese Academy of Sciences
2016-2018
University of Chinese Academy of Sciences
2016-2018
3D semantic occupancy prediction is a crucial task in visual perception, as it requires the simultaneous comprehension of both scene geometry and semantics. It plays role understanding scenes has great potential for various applications, such robotic vision perception autonomous driving. Many existing works utilize planar-based representations Bird's Eye View (BEV) Tri-Perspective (TPV). These aim to simplify complexity while preserving essential object information, thereby facilitating...
An experimental investigation of the breakup ${(\mathrm{C}{\mathrm{O}}_{2})}^{3+}$ induced by $\mathrm{N}{\mathrm{e}}^{4+}$ ion impact at incident energies 1.12 MeV was performed. By analyzing momentum distributions and kinetic three fragment ions, nonsequential sequential dissociation mechanisms are verified. In contrast to highly charged impact, two different decay pathways were observed in present experiment. One pathway originates from primary cation populated into...
Domain adaptation leverages labeled data from a source domain to learn an accurate classifier for unlabeled target domain. Since the collected in practical applications usually contain noise, weakly-supervised algorithm has attracted widespread attention researchers that tolerates with label noises or/and features noises. Several methods have been proposed mitigate difficulty of obtaining high-quality domains are highly related However, these assume obtain noise rate advance reduce negative...
Phases of quantum transition amplitudes contain very important information about physical processes. However, as a rule, they may not be accessed experimentally. This is, in particular, the case when collisions ions and atoms are studied. In this rapid communication, we explore double-electron capture $\ensuremath{\alpha}$ particles with CO molecules. Regarding atomic cores molecule two ``slits,'' show that ``double-slit'' interference can exploited to experimentally extract phases...
While decentralized training is attractive in multi-agent reinforcement learning (MARL) for its excellent scalability and robustness, inherent coordination challenges collaborative tasks result numerous interactions agents to learn good policies. To alleviate this problem, action advising methods make experienced share their knowledge about what do, while less strictly follow the received advice. However, method of sharing utilizing may hinder team's exploration better states, as can be...
While decentralized training is attractive in multi-agent reinforcement learning (MARL) for its excellent scalability and robustness, inherent coordination challenges collaborative tasks result numerous interactions agents to learn good policies. To alleviate this problem, action advising methods make experienced share their knowledge about what do, while less strictly follow the received advice. However, method of sharing utilizing may hinder team's exploration better states, as can be...
Tea picking has always been mainly manual, but with the development of technology, it possible to use computer image recognition technology assist robots in identifying tea leaves and them. However, special characteristics make difficult for machine recognize them high accuracy efficiency. In this paper, we compare currently available algorithms perform based on YOLOv5 algorithm model. Based original model, CBAM attention mechanism is added improve ability tea. The mAP our model can reach...