Lida Theodorou
Papers
2
Total Citations
35
H-Index
2
About
Lida Theodorou is a researcher advancing the frontier of few-shot object recognition, with a focus on enabling machines to learn new visual concepts from just a handful of examples. Her key contribution is the creation of the ORBIT dataset, a real-world, teachable object recognition benchmark that addresses a critical gap in computer vision: the reliance on massive, high-quality training sets. By designing ORBIT to capture the variability and noise of everyday user interactions—such as cluttered backgrounds, varied lighting, and object pose changes—Theodorou has provided a rigorous testbed for developing models that can learn on the fly, directly from end users. This work, which has garnered over 35 citations, is foundational for applications ranging from assistive robotics to personalized user interfaces, where adaptability is paramount. Her research not only pushes the boundaries of few-shot learning but also emphasizes practical, deployable AI that can be taught by non-experts. Theodorou’s contributions are a vital step toward more intuitive and accessible intelligent systems.
Research Focus
Key Achievements
Top Papers
- 1ORBIT: A Real-World Few-Shot Dataset for Teachable Object Recognition32 citations · 2021
- 2ORBIT: A Real-World Few-Shot Dataset for Teachable Object Recognition3 citations · 2021