Scholar
Alan Lukežič
Google Scholar ID: 1_5im9oAAAAJ
Mladi raziskovalec, Fakulteta za računalništvo in informatiko, Univerza v Ljubljani
Citations & Impact
Citations
6,074
H-index
20
i10-index
25
Publications
20
Co-authors
0
Contact
Publications
Resume (English only)
Research Experience
- Developing an open-source, low-cost, lightweight educational robotics platform with open software and hardware
- Researching adaptive deep perception methods for autonomous surface vehicles (USVs), focusing on general obstacle detection, long-term tracking with re-identification, implicit hazardous area detection, and multi-sensor fusion
- Developing autonomous navigation policies for mobile robots using deep reinforcement learning, trained entirely in simulation and deployed on real robots
- Designing novel deep architectures for visual surface defect inspection, capable of detecting large defects like cracks and smooth deformations such as dents on reflective surfaces
- Proposing D3S, a discriminative single-shot segmentation-based tracker that bridges visual object tracking and video object segmentation using two complementary geometric target models
- Developing maritime perception models using modalities beyond the visible spectrum (e.g., thermal imaging) for river navigation
- Building robust computer vision methods for USV navigation in uncontrolled environments, including monocular/stereo obstacle detection, efficient marine visual tracking, and sensor-fusion-based environment representation
- Designing long-term trackers with target re-detection capabilities for applications in surveillance, smart cameras, speaker-following in video conferencing, etc.
- Developing a Mask R-CNN-based method for large-scale traffic sign detection and recognition using the DFG dataset (7,000+ images, 200 traffic sign categories)
Co-authors