JC STEM Lab of MediaML @ HKU is founded in 2022. Our research focuses on endowing machines with the capability to perceive, understand, reconstruct, and interact with the visual world, with the following focuses:
label-efficient, transferrable and lifelong machine learning algorithms that are generalizable to various visual tasks
efficient compression techniques for visual data transmission
effective techniques for perceiving, interpreting and reasoning multi-modal (including vision, language, etc.) knowledge
learnable 3D models for extracting 3D semantics, inferring geometrical relationships, and rendering high-fidelity geometrical structures
deeply learned generative models for synthesizing or translating between multi-modal data.
JC STEM Lab of MediaML @ HKU is founded in 2022. Our research focuses on endowing machines with the capability to perceive, understand, reconstruct, and interact with the visual world, with the following focuses:
label-efficient, transferrable and lifelong machine learning algorithms that are generalizable to various visual tasks
efficient compression techniques for visual data transmission
effective techniques for perceiving, interpreting and reasoning multi-modal (including vision, language, etc.) knowledge
learnable 3D models for extracting 3D semantics, inferring geometrical relationships, and rendering high-fidelity geometrical structures
deeply learned generative models for synthesizing or translating between multi-modal data.
The JC STEM Lab of MediaML is run by Prof. Dong Xu (IEEE & IAPR Fellow, Member of the Academia Europaea, Clarivate Analytics 2018 Highly Cited Researcher in the field of Engineering, Receipt of T-MM 2014 Prize Paper Award and CVPR 2010 Best Student Paper Award [with the PhD students supervised by Prof. Xu]) at the University of Hong Kong, Hong Kong.
