- Augmented Reality Applications
- Image and Video Quality Assessment
- Virtual Reality Applications and Impacts
- Advanced Neural Network Applications
- Video Coding and Compression Technologies
- Interactive and Immersive Displays
- Advanced Data Compression Techniques
- Caching and Content Delivery
- Advanced Antenna and Metasurface Technologies
- Context-Aware Activity Recognition Systems
- Internet of Things and Social Network Interactions
- Tactile and Sensory Interactions
- Video Analysis and Summarization
- Machine Learning in Healthcare
- Spectroscopy and Chemometric Analyses
- Engineering Applied Research
- Multimedia Communication and Technology
- IoT and Edge/Fog Computing
- Spectroscopy Techniques in Biomedical and Chemical Research
- Innovative Educational Techniques
- Sharing Economy and Platforms
- Privacy, Security, and Data Protection
- Stochastic Gradient Optimization Techniques
- Smart Agriculture and AI
- Video Surveillance and Tracking Methods
Xi'an Jiaotong University
2019-2024
City University of Hong Kong
2022-2024
Southeast University
2020
China Agricultural University
2019
With the rapid development of mobile devices and deep learning, smart applications using learning technology have sprung up. It satisfies multiple needs users, network operators service providers, rapidly becomes a main research focus. In recent years, has achieved tremendous success in image processing, natural language analysis other fields. Despite task performance been greatly improved, resources required to run these models increased significantly. This poses major challenge for...
Avatar-mediated symmetric Augmented Reality (AR) telepresence has emerged with the ability to empower users located in different remote spaces interact each other 3D through avatars. However, have heterogeneous structures and features, which bring difficulties synchronizing avatar motions real user adapting local scenes. To overcome these issues, existing methods generate mutual movable or retarget placement of limit experience a small sub-area space, fix positions avatars, adjust...
Safety helmet is an important protective tool for workers in the industrial production, and clothing worn by can play a role distinguishing other person. To solve problem of non-contact on-line monitoring to wear safety as required, this paper proposes improved-YoloV3 method intelligent detection identification algorithm. The large-size input layer added multi-scale prediction size anchor boxes adjusted enhance ability small-size clothing. algorithm not only meets real-time supervision...
Augmented Reality (AR) teleconferencing allows spatially distributed users to interact with each other in 3D through agents their own physical environments. Existing methods leveraging volumetric capturing and reconstruction can provide a high-fidelity experience but are often too complex expensive for everyday use. Other solutions target mobile effortless-to-setup on AR Head Mounted Displays (HMD). They directly transplant the conventional video conferencing onto an AR-HMD platform or use...
To further enhance the immersion perception of remote interaction, avatars can be involved harnessing Head Mounted Display (HMD) based Augmented Reality (AR). In our demonstration, we present an avatar interaction system AvatarMeeting, enabling users to meet with peers through interactive personalized just like face face. Specifically, propose a novel framework including consumer-grade set-up, complete transmission scheme and processing pipeline, which consists prescan modeling, pose...
It is extremely important to correctly identify the cultivars of maize seeds in breeding process maize. In this paper, transfer learning as a method deep adopted establish model by combining with hyperspectral imaging technology. The haploid can be recognized from large amount diploid ones great accuracy through model. First, information on each wave band collected using technology, and then recognition built VGG-19 network, which pretrained large-scale computer vision database (Image-Net)....
As the scale of models and training data continues to grow, there is an expanding reliance on more GPUs train large-scale models, which inevitably increases likelihood encountering dynamic stragglers that some devices lag behind in performance occasionally. However, hybrid parallel training, one de facto paradigms large typically sensitive stragglers. This paper presents Malleus, a straggler-resilient framework for models. Malleus captures straggler issues at nuanced, per-GPU granularity...
Augmented Reality (AR) teleconferencing allows separately located users to interact with each other in 3D through agents their own physical environments. Existing methods leveraging volumetric capturing and reconstruction can provide a high-fidelity experience but are often too complex expensive for everyday usage. Other solutions target mobile effortless-to-setup on AR Head Mounted Displays (HMD). They directly transplant the conventional video conferencing onto an AR-HMD platform or use...
Augmented Reality (AR) teleconferencing aims to enable remotely separated users meet with each other in their own physical spaces as if they are face-to-face. Among all solutions, the video-avatar-based approach has advantage of balancing fidelity and sense co-presence using easy-to-setup devices, including only a camera an AR Head-Mounted Display (HMD). However, non-verbal cues indicating "who is looking at whom" always lost or misdelivered multiparty experiences. To make aware such cues,...
HTTP adaptive streaming (HAS) is the state-of-the-art technology for improving quality of user experience under conditions time-varying available bandwidth. Developing bitrate adaptation algorithm becomes more challenging with transition to 4G cellular networks. The features bandwidth variation caused by changes in radio channel are significantly different from pattern resources. In this paper, we propose a networks differentiation. By investigating data profiles collected an actual network...
Mobile video applications are becoming increasingly prevalent and enriching the way people learn entertained. However, on mobile terminals with inherently limited resources, streaming services consume too much energy bandwidth, which is an urgent problem to solve. At present, research cost-effective typically focuses management of data transmission. Among such studies, some new approaches consider user's behavior further optimize these studies have not adequately discussed specific impact...
A plethora of mobile video applications involving all aspects social life are becoming increasingly prevalent. However, for resource-hungry and delay-sensitive multi-view such as multi-channel conferencing 3D videos, the hardware resources terminals becomes a bottleneck in concurrently decoding multiple videos. Fusing views into single-view stream cloud before transmission can unload computation terminals. But delay increment caused by fusing this cloudbased streaming makes strategy...
This research employs the iPhone as a case study and conducts survey to investigate consumers' willingness-to-pay (WTP) for privacy. We discover that consumers not only value their privacy in digital realm but are also willing invest considerably privacy-protection services. What makes our findings particularly intriguing is relationship between habits willingness-to-pay. Surprisingly, individuals who may exhibit lower vigilance safeguarding online paradoxically demonstrate higher services...
Optimizing the Quality of Experience (QoE) mobile video streaming service has emerged as a hot research topic. Simulation been dominant methodology validation. However, it is becoming clear that simulation does not sufficiently capture network capacity irregularity real network, especially with mobility. The researchers tend to evaluate their optimizations using prototype implementations by world experiments. Instead directly running time and cost intensive field test, more efficient on...
Carbon fiber is an excellent functional and structural material with combining a variety of properties. In order to combat photoelectronic reconnaissance guidance weapons, carbon widely used in the military field as interference millimeter wave smoke agent, has become hot issue screen passive technology. explore attenuation performance compare effect powder short-cut on wave, experiments have measured particle sizes 25, 35, 70, 140, 200, 400, 800 chopped 3 waves 8 waves. The research results...