A5101622670- Reinforcement Learning in Robotics
- Artificial Intelligence in Games
- Remote-Sensing Image Classification
- Remote Sensing and LiDAR Applications
- Evolutionary Algorithms and Applications
- Video Analysis and Summarization
- Advanced Neural Network Applications
- Remote Sensing in Agriculture
- Anomaly Detection Techniques and Applications
- Advanced Multi-Objective Optimization Algorithms
- Video Surveillance and Tracking Methods
- Advanced Bandit Algorithms Research
- Smart Agriculture and AI
- Human Pose and Action Recognition
China University of Geosciences
2024
Chinese Academy of Sciences
2022
University of Chinese Academy of Sciences
2021-2022
Shenyang Institute of Automation
2021-2022
Nanjing Agricultural University
2021
Semantic segmentation of remote sensing images plays a critical role in areas such as urban change detection, environmental protection, and geohazard identification. Convolutional Neural Networks (CNN) have been excessively employed for semantic over the past few years; however, limitation CNN is that there exists challenge extracting global context images, which vital segmentation, due to locality convolution operation. It informed recently developed Transformer equipped with powerful...
Single-modal images carry limited information for features representation, and RGB fail to detect grass weeds in wheat fields because of their similarity shape. We propose a framework based on multi-modal fusion accurate detection natural environment, overcoming the limitation single modality detection. Firstly, we recode single-channel depth image into new three-channel like structure image, which is suitable feature extraction convolutional neural network (CNN). Secondly, multi-scale...
Video recognition remains an open challenge, requiring the identification of diverse content categories within videos. Mainstream approaches often perform flat classification, overlooking intrinsic hierarchical structure relating categories. To address this, we formalize novel task video recognition, and propose a video-language learning framework tailored for recognition. Specifically, our encodes dependencies between category levels, applies top-down constraint to filter predictions. We...
Reinforcement learning has successfully been used in many applications and achieved prodigious performance (such as video games), DQN is a well-known algorithm RL. However, there are some disadvantages practical applications, the exploration exploitation dilemma one of them. To solve this problem, common strategies about like ɛ–greedy have risen. Unfortunately, sample inefficient ineffective because uncertainty later exploration. In paper, we propose model-based method that learns state...
Model-free reinforcement learning methods have successfully been applied to practical applications such as decision-making problems in Atari games. However, these inherent shortcomings, a high variance and low sample efficiency. To improve the policy performance efficiency of model-free learning, we propose proximal optimization with model-based (PPOMM), fusion method both learning. PPOMM not only considers information past experience but also prediction future state. adds next state...