A5103177111- Fire Detection and Safety Systems
- Video Surveillance and Tracking Methods
- IoT-based Smart Home Systems
- Fire dynamics and safety research
- Antenna Design and Analysis
- Energy Harvesting in Wireless Networks
- Image Enhancement Techniques
- Advanced Image Processing Techniques
- COVID-19 diagnosis using AI
- Full-Duplex Wireless Communications
- Multimodal Machine Learning Applications
- Domain Adaptation and Few-Shot Learning
- Evacuation and Crowd Dynamics
- Advanced Vision and Imaging
Zhejiang University
2023-2024
Ningbo University
2023-2024
Wuhan University
2024
The current irregularities in existing public Fire and Smoke Detection (FSD) datasets have become a bottleneck the advancement of FSD technology. Upon in-depth analysis, we identify core issue as lack standardized dataset construction, uniform evaluation systems, clear performance benchmarks. To address this drive innovation technology, systematically gather diverse resources from sources to create more comprehensive refined benchmark. Additionally, recognizing inadequate coverage scenes,...
An effective Fire and Smoke Detection (FSD) analysis system is of paramount importance due to the destructive potential fire disasters. However, many existing FSD methods directly employ generic object detection techniques without considering transparency smoke, which leads imprecise localization reduces performance. To address this issue, a new Attentive Model (a-FSDM) proposed. This model not only retains robust feature extraction fusion capabilities conventional algorithms but also...
Portrait shadow removal is a challenging task due to the complex surface of face. Although existing work in this field makes substantial progress, these methods tend overlook information background areas. However, not only contains some important illumination cues but also plays pivotal role achieving lighting harmony between face and after elimination. In paper, we propose Context-aware Illumination Restoration Network (CIRNet) for portrait removal. Our CIRNet consists three stages. First,...
Source-Free domain adaptive Semantic Segmentation (SFSS) aims to transfer knowledge from source the target with only pre-trained segmentation model and unlabeled dataset. Only a few works have been researched for SFSS, relying on entropy minimization, pseudo-labeling. Nevertheless, due bias, these methods tend suffering confusion of classes similar visual appearance in different domains. To address above issue, we propose enhance discriminability towards samples masked image modeling spatial...