A5101581314- Advanced Neural Network Applications
- Power Line Inspection Robots
- Vehicle License Plate Recognition
- Multimodal Machine Learning Applications
- Remote Sensing and LiDAR Applications
- Advanced Image and Video Retrieval Techniques
- Traffic control and management
- Autonomous Vehicle Technology and Safety
- Traffic Prediction and Management Techniques
- Video Surveillance and Tracking Methods
- Domain Adaptation and Few-Shot Learning
Zhengzhou University
2023
National University of Defense Technology
2019
Reinforcement learning (RL)-based traffic signal control has been proven to have great potential in alleviating congestion. The state definition, which is a key element RL-based control, plays vital role. However, the data used for definition literature are either coarse or difficult measure directly using prevailing detection systems control. This paper proposes deep reinforcement learning-based method uses high-resolution event-based data, aiming achieve cost-effective and efficient...
Power lines are critical infrastructure components in power grid systems. Strand breakage is a kind of serious defect that can directly impact the reliability and safety supply. Due to slender morphology difficulty acquiring sufficient sample data, strand detection remains challenging task. Moreover, corporations prefer detect these defects on-site during line inspection using unmanned aerial vehicles (UAVs), rather than transmitting all data central server for offline processing which...
Low computational complexity and high segmentation accuracy are both essential to the real-world semantic tasks. However, speed up model inference, most existing approaches tend design light-weight networks with a very limited number of parameters, leading considerable degradation in due decrease representation ability networks. To solve problem, this paper proposes novel method improve capacity obtaining information for network. Specifically, feature refinement module (FRM) is proposed...
In the published work [...]
Semantic segmentation of nighttime images plays an equally important role as that daytime in autonomous driving, but the former is much more challenging due to poor illuminations and arduous human annotations. In this paper, we propose a novel domain adaptation network (DANNet) for semantic without using labeled image data. It employs adversarial training with dataset unlabeled contains coarsely aligned day-night pairs. Specifically, pairs, use pixel-level predictions static object...