Akshaya Thippur
Papers
4
Total Citations
68
H-Index
4
About
Akshaya Thippur’s research lies at the intersection of computer vision and robotics, with a central focus on enabling machines to achieve robust scene understanding. Her work pioneers the integration of top-down spatial reasoning with bottom-up object class recognition, demonstrating that a robot’s perception is significantly enhanced when it considers not just what an object looks like, but where it is located in relation to other objects. This foundational insight, detailed in her most-cited paper (35 citations), shows that combining spatial context with visual features dramatically improves recognition accuracy. Thippur has also made key contributions to modelling spatial relations, comparing qualitative and metric approaches to infer object labels in real-world scenes. Her research extends to the challenging domain of hand pose estimation, where she has evaluated visual shape features for applications in sign language recognition and robot learning from demonstration. To support long-term autonomous learning, she introduced the KTH-3D-TOTAL dataset, a resource designed to help robots discover spatial structures and generalize across varying object instances and scenes over time. Through her work, Thippur has advanced the field by showing that true scene understanding requires a holistic, context-aware approach.
Research Focus
Key Achievements
Top Papers
- 1
- 2
- 3Inferring hand pose: A comparative study of visual shape features14 citations · 2013
- 4