Full Publications & Under review
Journal/ Conference/ project/
My research interest lies at the computer network and systems, with intersection over multimedia, HCI, and robot learning.
Description: The next generation of real-time interactive multimedia, such as metaverse, virtual reality, and videoconferencing, requires much more bandwidth and computation than current applications. However, the underlying system and algorithm design is not sufficient to provide a satisfactory, low latency experience for multiple users. In my research, I have three goals: (1) Networked VR system: To overcome the hardware limitations of current devices, I am working on building a next-generation streaming system for mobile augmented/mixed reality, multi-user extended reality, and the metaverse that can be supported on commodity mobile devices. (2) ML for networked systems: Exploiting new perspective of applying machine learning techniques to improve the performance and reliability of large-scale networked systems. (3) Video Streaming and Compression: For emerging 360-degree /holographic/ volumetric video, we are researching ways to stream them in real-time, focusing on the networking challenges that come with it. We are also using deep neural compression to efficiently transmit high-dimensional video data and display high-quality 3D content on devices with limited resources. (4) Spatial Computing: By understanding and perceiving the dynamics of the multimodal world and human priors, we use optimization techniques such as on-device and edge-assisted collaborative caching and streaming to reduce overhead. We are also working on providing accurate and robust tracking of multiple agents for various tasks and generating their avatars for multi-user XR.

A Networking Perspective of Volumetric Videos: Opportunities and Challenges
Other author
Submitted to Magazine, 2022. [arXiv] [Project Page] [Code]
Ebublio: Edge Assisted Multi-user 360-Degree Video Streaming
Yili Jin, Junhua Liu, Fangxin Wang, Shuguang Cui,
Expansion of conference version. Minor Revision (JCR Q1), 2022. [arXiv] [Project Page] [Code]
Melt: Streaming Neural Fields-Enhanced Volumetric Video on Mobile Architectures.
First author,
Expansion of conference version, 2023. [arXiv] [Project Page] [Code]
Neural Fields for Networking: Advances, Opportunities and Outlook
First author,
Manuscript and arXiv coming soon, 2023. [arXiv] [Project Page] [Code]
FSVVD: A Dataset of Full Scene Volumetric Video
Kaiyuan Hu, Yili Jin, Haowen Yang, Junhua Liu, Fangxin Wang,
In submission, 2023. [arXiv] [Project Page] [Code]
Melt: Implicit Neural Representation-Enhanced Volumetric Video Streaming on Mobile Devices
Junhua Liu, Yuanyuan Wang, Yan Wang, Fangxin Wang,
In submission, 2023. [arXiv] [Project Page] [Code]
CaV3: Cache-assisted Viewport Adaptive Volumetric Video Streaming
Junhua Liu, Boxiang Zhu, Fangxin Wang, Yili Jin, Wenyi Zhang, Zihan Xu, Shuguang Cui,
Conditionally accepted for presentation and publication at (IEEE VR), 2023. [arXiv] [Project Page] [Code]
Efficient NeRF: Utilizing Efficient Neural Radiance Fields for Downstream Applications
Junhua Liu, Biaolin Wen, Rui He, Yuanyuan Wang, Yan Wang, Hongwei Qin,
In submission, 2022. [PDF] [Project Page]
Where Are You Looking? A Large-Scale Dataset of Head and Gaze Behavior for 360-Degree Videos and a Pilot Study.
Yili Jin*, Junhua Liu*, Fangxin Wang, Shuguang Cui,
ACM Multimedia (MM), 2022. [arXiv] [Project Page] [Code]
Viewport-Aware Adaptive Volumetric Video Streaming
Zihan Xu, Wenyi Zhang, Junhua Liu, Yili Jin, Fangxin Wang, Lian Zhao, Shuguang Cui,
In submission, 2022. [arXiv] [Project Page] [Code]
Ebublio: Edge Assisted Multi-user 360-Degree Video Streaming
Yili Jin, Junhua Liu, Fangxin Wang, Shuguang Cui,
IEEE Conference on Virtual Reality and User Interfaces (IEEE VR), 2022. (Short paper) [arXiv] [Project Page] [Code]
Improving the Generation Ability of Stock-trading Agents With Generated Samples
Biaolin Wen, Junhua Liu, Tianshu Yu, Bowen Zhang,
In submission, 2022. [arXiv] [Project Page] [Code]
NeRS: A network-based Reinforced system for Embodied Intelligence
Junhua Liu,
Project on DDA4230: Reinforcement Learning, 2022. [arXiv] [Project Page] [Code]