Piotr Ozimek
Papers
2
Total Citations
14
H-Index
2
About
Piotr Ozimek is a researcher at the forefront of biologically inspired computer vision, specializing in egocentric perception and data-efficient deep learning. His work bridges neuroscience and artificial intelligence by modeling the mammalian retino-cortical visual pathway to create software retinas that dramatically reduce visual data while preserving scale and rotation invariance. In his seminal 2019 paper, "A Space-Variant Visual Pathway Model for Data Efficient Deep Learning" (8 citations), Ozimek demonstrated how adopting biological retino-cortical mapping can significantly improve the efficiency of deep convolutional neural networks for robot vision systems. His earlier foundational work, "Egocentric Perception using a Biologically Inspired Software Retina Integrated with a Deep CNN" (6 citations), presented at the first EPIC workshop in Amsterdam, established the concept of a software retina capable of substantial visual data reduction. These contributions are particularly impactful for autonomous systems requiring real-time visual processing with limited computational resources. Ozimek's research offers a promising path toward more efficient, nature-inspired AI that could enable next-generation wearable and robotic perception systems.
Research Focus
Key Achievements
Top Papers
- 1A Space-Variant Visual Pathway Model for Data Efficient Deep Learning8 citations · 2019
- 2