Aleksandra Kos
Papers
2
Total Citations
7
H-Index
2
About
Aleksandra Kos is a computer vision researcher whose work focuses on the critical challenge of detecting small and tiny objects in high-resolution imagery. Her primary contributions lie in developing enhanced lightweight detection frameworks that balance computational efficiency with accuracy. Kos’s most influential work introduces an innovative object tracking-based region of interest (ROI) proposal method, which intelligently narrows the search space for small objects, dramatically improving detection performance without the heavy computational cost of traditional sliding-window approaches. This technique is particularly valuable for applications like drone surveillance, satellite imagery analysis, and autonomous systems where small objects are often missed. Her two most-cited papers, published in 2024 and 2025, have together accumulated 7 citations, establishing her as an emerging voice in efficient object detection. Kos’s research addresses a persistent bottleneck in computer vision: the trade-off between model lightness and detection sensitivity for tiny targets. By integrating temporal tracking cues into the ROI selection process, she has created a practical solution that pushes the boundaries of what lightweight models can achieve, making her work highly relevant for real-time, resource-constrained environments.
Research Focus
Key Achievements
Top Papers
- 1
- 2