Papers
6
Total Citations
304
H-Index
5
About
Alexey Abramov’s research lies at the intersection of computer vision, robotics, and real-time GPU computing, with a focus on enabling robots to understand and replicate human actions through visual observation. His major contributions include developing novel representations for object-action relations, such as semantic scene graphs that allow robots to recognize manipulations and transfer them for execution. This work, particularly his 2011 paper on learning the semantics of object–action relations by observation, has garnered 180 citations, underscoring its impact on the field of robot learning from demonstration. Abramov also advanced real-time stereo video segmentation on mobile GPUs, achieving frame rates of 23 Hz for regular videos—a critical achievement for resource-constrained robotic platforms. His 2012 paper on this topic, with 22 citations, highlights his ability to bridge algorithmic innovation with practical deployment. Additionally, he pioneered self-supervised on-line training for object recognition, reducing the need for human supervision in robotic learning. Abramov’s work has been recognized for its efficiency and scalability, making him a notable figure in autonomous systems and visual perception.
Research Focus
Key Achievements
Top Papers
- 1Learning the semantics of object–action relations by observation180 citations · 2011
- 2Categorizing object-action relations from semantic scene graphs80 citations · 2010
- 3
- 4
- 5
- 6Occlusion Handling in Video Segmentation via Predictive Feedback3 citations · 2012