Jinfa Huang
Papers
2
Total Citations
13
H-Index
2
About
Jinfa Huang is a researcher advancing the frontier of multimodal AI and semantic scene understanding. His work centers on developing deep learning models that bridge the gap between visual perception and linguistic reasoning, with a particular focus on dynamic environments and social intelligence. In his highly cited 2023 paper, "Cross-Modality Time-Variant Relation Learning for Generating Dynamic Scene Graphs," Huang introduced a novel framework for constructing evolving scene graphs from video. This work, which has garnered 11 citations, enables machines to track how objects and their relationships change over time, directly addressing critical challenges in autonomous navigation and robotic task planning. Earlier, Huang explored the nuanced domain of social AI with "LDNN: Linguistic Knowledge Injectable Deep Neural Network for Group Cohesiveness Understanding." This 2020 paper pioneered a method for embedding linguistic cues into neural networks to interpret the subtle, complex dynamics of human group intimacy—a capability essential for developing more empathetic dialogue robots. By tackling both the physical and social dimensions of visual understanding, Huang is helping to build AI systems that can perceive not just what is happening, but how people are relating to one another.
Research Focus
Key Achievements
Top Papers
- 1
- 2