- Image and Video Quality Assessment
- Visual Attention and Saliency Detection
- Video Coding and Compression Technologies
- Video Surveillance and Tracking Methods
- IoT and Edge/Fog Computing
- Mobile Crowdsensing and Crowdsourcing
- Target Tracking and Data Fusion in Sensor Networks
- Gaze Tracking and Assistive Technology
- Markov Chains and Monte Carlo Methods
- Distributed Sensor Networks and Detection Algorithms
- Indoor and Outdoor Localization Technologies
- Digital Media Forensic Detection
- Privacy-Preserving Technologies in Data
- Network Security and Intrusion Detection
- Telecommunications and Broadcasting Technologies
- Human Mobility and Location-Based Analysis
- Adversarial Robustness in Machine Learning
- Blind Source Separation Techniques
- Green IT and Sustainability
- Advanced Data Compression Techniques
- Advanced Steganography and Watermarking Techniques
- Advanced Data Storage Technologies
- Computer Graphics and Visualization Techniques
- Advanced Computing and Algorithms
- Smart Grid Security and Resilience
Simon Fraser University
2021-2025
Beijing University of Posts and Telecommunications
2019
IBM Research (China)
2012
Beihang University
2008-2012
The widespread integration of the Internet Things with sensors like depth-of-field cameras, LiDAR scanners, and eye-tracking infrared sensors, in head-mounted devices, has ushered a new era immersive digital experiences. Full-scene volumetric video (VV), key innovation this integration, provides deeply experience by capturing richness detail 3D world. However, its massive data volume presents significant streaming challenges. While tile-based viewport approaches have been proposed, they...
The fifth generation (5G) communication systems have seen initial success in boosting a broad spectrum of mobile networked applications. However, emerging applications, notably immersive eXtended Reality (XR), already posed significant new challenges to today's 5G given their ultra-high expectations on data rate and latency. They also demand deep integration computation for analytics. In particular, by analyzing the vision captured devices, Mobile Vision Analytics (MVA) facilitates...
Immersive full-scene volumetric video (VV) showcases the richness and detail of 3D world, yet poses significant streaming challenges given its massive data volume. Existing tile-based viewport approaches struggle to effectively adapt VV owing their small buffer limitation, high tile segmentation overhead, lack consideration.
With the reduced hardware costs of omnidirectional cameras and proliferation various extended reality applications, more 360° videos are being captured. To fully unleash their potential, advanced video analytics is expected to extract actionable insights situational knowledge without blind spots from videos. In this paper, we present OmniSense, a novel edge-assisted framework for online immersive analytics. OmniSense achieves both low latency high accuracy, combating significant computation...
Deep learning networks are widely used in various systems that require classification. However, deep vulnerable to adversarial attacks. The study on attacks plays an important role def... | Find, read and cite all the research you need Tech Science Press
360° videos are gaining popularity, but immersive analytics, particularly in object detection, confront challenges from complex scenes and high data volume. This imposes significant burdens on individual users resource-limited edge devices. Fortunately, Machine Learning as a Service (MLaaS) offers an economical solution for quick deployment without specific hardware or expertise. However, current MLaaS mostly 2D image-designated not optimized the distinctive characteristics of raw video...
In the 5G era (and upcoming 6G), mobile edge computing (MEC) has been advocated to serve massive amount of Internet Things (IoT) devices by base stations (BSs) and data centers (EDCs). Geo-distributed EDCs are generally much smaller scales as compared mega hence lower costs, but can have fast response their users so satisfy demands real-time applications. As reliability availability heavily depend on electrical power supply, most equipped with battery groups backup in case grid load shedding...
Today's electrical grid is experiencing a fast transition toward smart infrastructure. Modern expected to integrate Artificial Intelligence of Things (AIoT)-empowered energy management systems (EMS) sense, analyze, and optimize the power consumption QoS diverse end users. Non-Intrusive Load Monitoring (NILM) plays key role in this transition, particularly considering that many legacy devices/appliances may not have built-in sensors. Yet most NILM solutions rely on large (often impractical)...
Data centers are essential components in the current digital world. The number and scales of data have both increased a lot recent years. distributed standing out as promising solution due to development modern applications which need massive amount computation resource strict response requirement. However, compared centralized centers, more fragile when power supply is unstable. Power constraints or outages because electrical load shedding other reasons will significantly affect service...
Distributed systems, especially those providing cloud services, endeavor to construct sufficiently reliable storage in order attract more customers. Generally, pure replication and erasure code are widely adopted distributed systems guarantee data storage, yet both of them contain some deficiencies. Pure consumes too much extra bandwidth, while seems not so high-efficiency only suitable for read-only context. The authors proposed REPERA as a hybrid mechanism combining leverage their...
360° videos are becoming one of the major media in recent years, providing immersive experience for viewers with more interactions compared to traditional videos. Most today's implementations rely on bulky Head-Mounted Displays (HMDs) or require touch screen operations interactive display, which not only expensive but also inconvenient viewers. In this paper, we demonstrate that video streaming can be done hints from gaze movement detected by front camera mobile devices (e.g., a smartphone)....
Although it is known that exact sampling algorithm easy to construct and less sensitive noise, the samples distribution of deviates from target states due local dependent coupling problem. A new algorithm, named with directional threshold (ES-DT) introduced. The main advantage in comparison traditional can control a rejection strategy Markov chain during path growth, closely approach ideal based on maintaining density. Simulation experiments show effectiveness proposed algorithm.
This paper proposed a new algorithm of Coupling from the Past (CFTP) with directional threshold. CFTP, also called Exact Sampling, established in 1996 by Propp and Wilson, aimed that it would eliminate need to compute Markov chain convergence rate for quality control. CFTP was used mixture models Monte Carlo worked well low computation complexity problems. The appealing its invariant structure but many applications process coupling not always an independent process. And this suboptimal...
The uncertainty of the sensors position will render measurements useless in many scenarios. To track a maneuvering target sensor network which nodes are unknown locations, states must be estimated jointly with positions. In this paper number is given, from independent and there no beacon available for aid node locating. We extended frame work Joint probabilistic data association (JPDA), an approach multi-target tracking known targets, adapted it our problem obtained fused results by weights...