A5073936626- Anomaly Detection Techniques and Applications
- Video Surveillance and Tracking Methods
- Concrete and Cement Materials Research
- Structural Behavior of Reinforced Concrete
- Innovative concrete reinforcement materials
- Emotion and Mood Recognition
- Parental Involvement in Education
- Face recognition and analysis
- Adversarial Robustness in Machine Learning
- Structural Load-Bearing Analysis
- Cognitive and developmental aspects of mathematical skills
- Human Mobility and Location-Based Analysis
- Face and Expression Recognition
- Magnesium Oxide Properties and Applications
- Advanced Computing and Algorithms
- Structural Engineering and Vibration Analysis
- Innovative Teaching and Learning Methods
- Digital Media Forensic Detection
- Data Stream Mining Techniques
University of British Columbia
2021-2023
Fujian Electric Power Survey & Design Institute
2020
Recent studies show that small perturbations in video frames could misguide single object trackers. However, such attacks have been mainly designed for digital-domain videos (i.e., perturbation on full images), which makes them practically infeasible to evaluate the adversarial vulnerability of trackers real-world scenarios. Here we made first step towards physically feasible against visual tracking real scenes with a universal patch camouflage Fundamentally different from physical...
Low carbon and high performance have become key trends in the development of construction materials. Understanding mechanism by which curing conditions affect mechanical properties high-ductility geopolymer concrete (HDGC) is significant importance. This study investigated three sealing temperatures (room temperature, 45 °C, 60 °C) four durations (1 day, 3 days, 5 7 days), while considering two final ages (7 days 28 days) to explore their effects on axial tensile compressive HDGC. The...
One promising way to improve the performance of a small deep network is knowledge distillation. Performances smaller student models with fewer parameters and lower computational cost can be comparable that larger teacher in specific computer vision tasks. Knowledge distillation especially attractive for high-accuracy real-time crowd counting task our daily lives, where resource limited model efficiency extremely important. In this paper, we propose novel task-specific framework counting,...
Facial action units (FAUs) are critical for fine-grained facial expression analysis. Although FAU detection has been actively studied using ideally high quality images, it was not thoroughly under heavily occluded conditions. In this paper, we propose the first occlusion-robust recognition method to maintain performance heavy occlusions. Our novel approach takes advantage of rich information from latent space masked autoencoder (MAE) and transforms into features. Bypassing occlusion...
Facial action units (FAUs) are critical for fine-grained facial expression analysis. Although FAU detection has been actively studied using ideally high quality images, it was not thoroughly under heavily occluded conditions. In this paper, we propose the first occlusion-robust recognition method to maintain performance heavy occlusions. Our novel approach takes advantage of rich information from latent space masked autoencoder (MAE) and transforms into features. Bypassing occlusion...