Spyridon Thermos
Papers
1
Total Citations
7
H-Index
1
About
Spyridon Thermos is a researcher advancing the frontiers of computer vision and robotics through a focus on human-object interaction and scene understanding. His work centers on the challenging task of inferring object affordances—the actionable properties and interaction possibilities of objects—directly from visual data. In his highly cited 2021 paper, "Joint Object Affordance Reasoning and Segmentation in RGB-D Videos," Thermos introduced a novel framework that simultaneously reasons about an object’s functional parts and segments them in dynamic video streams. This integrated approach represents a significant leap beyond static image analysis, enabling robots and AI systems to interpret not just what an object is, but how it can be used in real-world, temporal contexts. By leveraging RGB-D video, his method captures both spatial and depth cues, paving the way for more intuitive and safe human-robot collaboration. Though early in his career, with his most impactful work already garnering 7 citations, Thermos is establishing himself as a key voice in embodied AI. His research directly addresses a core bottleneck in autonomous systems: the ability to perceive and act upon the environment with human-like understanding.
Research Focus
Key Achievements
Top Papers
- 1Joint Object Affordance Reasoning and Segmentation in RGB-D Videos7 citations · 2021